0:00:10 [MUSIC]
0:00:13 Hello, and welcome to the NVIDIA AI podcast.
0:00:15 I’m your host, Noah Kravitz.
0:00:19 90% of the information transmitted to human brains is visual.
0:00:22 So while advances related to large language models and
0:00:26 other language processing technology have pushed the frontier of AI forward in
0:00:28 a hurry over the past few years.
0:00:32 Visual information is integral for AI to act with the physical world,
0:00:34 which is where computer vision comes in.
0:00:37 Roboflow empowers developers of all skill sets and
0:00:41 experience levels to build their own computer vision applications.
0:00:44 The company’s platform addresses the universal pain points developers face
0:00:47 when building CV models, data management to deployment.
0:00:51 Roboflow is currently used by over 16,000 organizations and
0:00:55 half the Fortune 100, totaling over 1 million developers.
0:00:59 And they’re a member of NVIDIA’s Inception program for startups.
0:01:03 Roboflow co-founder and CEO, Joseph Nelson, is with us today to talk about
0:01:08 his company’s mission to transform industries by democratizing computer vision.
0:01:09 So let’s jump into it.
0:01:13 Joseph, welcome, and thank you for joining the NVIDIA AI podcast.
0:01:14 >> Thanks so much for having me.
0:01:19 >> I’m excited to talk CV, there’s nothing against language, love language.
0:01:22 But there’s been a lot of language stuff lately, which is great.
0:01:25 But I’m excited to hear about Roboflow.
0:01:26 So let’s jump into it.
0:01:30 Maybe you can start by talking a little bit more about your mission and
0:01:34 what democratizing computer vision means and making the world programmable.
0:01:37 >> At the highest level, as you just described,
0:01:40 the vast majority of information that humans process happens to be visual
0:01:41 information.
0:01:45 In fact, I mean, humans, we had our sense of sight before we even created language.
0:01:46 It’s how we understand the world.
0:01:50 It’s how we understand the things around us, it’s how we engage with the world.
0:01:56 And because of that, I think there’s this massive untapped potential
0:01:59 to have technology and systems have visual understanding to
0:02:04 a similar way that humans do all across, really, we say the universe.
0:02:07 So when we say our North Star is to make the world programmable,
0:02:09 what we really mean is that any scene and
0:02:13 anything will have software that understands it.
0:02:14 And when you have software that understands something,
0:02:16 you can improve that system.
0:02:18 You can make it be more efficient, you can make it be more entertaining,
0:02:20 you can make it be more engaging.
0:02:23 I mean, at Roboflow, we’ve seen folks build things from understanding
0:02:27 cell populations under a microscope all the way to discovering new galaxies
0:02:28 through a telescope.
0:02:31 And everything in between is where vision and video and
0:02:33 understanding comes into play.
0:02:36 So if this AI revolution is to reach its full potential,
0:02:38 it needs to make contact with the real world.
0:02:42 And it turns out the real world is one that is very visually rich and
0:02:43 needs to be understood.
0:02:47 So we build the tools, the platform, and the community to accelerate that transition.
0:02:51 So maybe you can tell us a little bit about the platform and
0:02:55 kind of within that how your mission and North Star kind of shape the way you
0:02:57 develop products and build out user experiences.
0:03:02 And I should mention a great shout out from Jensen in the CES keynote earlier
0:03:04 this year for you guys.
0:03:06 And you raised Series B late last year.
0:03:10 So I want to congratulate you on that as well before I forget.
0:03:10 >> I appreciate it.
0:03:12 Yeah, it’s a good start.
0:03:15 I mean, as you said, a million developers, but there’s 70 million developers out there.
0:03:18 There’s billions that will benefit from having visual understanding.
0:03:21 I mean, in fact, in that Jensen shout out, just like maybe the sentence or
0:03:25 two before he described some of the visual partners that are fortunate to work
0:03:26 with the NVIDIA team.
0:03:30 He described that the way NVIDIA sees it is that global GDP is $100 trillion,
0:03:35 and he describes that visual understanding is like $50 trillion of that opportunity.
0:03:39 So basically half of all global GDP is predicated on these
0:03:43 operationally intensive, visual, centric, autonomy-based sorts of use cases.
0:03:47 And so, the level of impact that visual understanding will have in the world is
0:03:50 just a fraction of what it will look like as we progress.
0:03:53 Now, in terms of like how we think about doing that, so
0:03:56 it’s really about empowering the builders and giving the certainty and
0:03:58 capability to the enterprise.
0:04:02 So for example, anyone that’s building a system for visual understanding often
0:04:07 needs to have some form of visual input, camera, video, something like this.
0:04:08 >> Sure.
0:04:10 >> You need to have a model because the model is going to help you act,
0:04:15 understand, and react to whatever maybe actual insight you want to understand.
0:04:17 And then you want to run that model some more, you want to deploy it.
0:04:21 And commonly, you even want to chain together models or
0:04:23 have a system that triggers some alerts or
0:04:26 some results based on information that it’s understanding.
0:04:29 So Robofill provides the building blocks, the platform, and
0:04:34 the solutions so that over a million developers and half the Fortune 100
0:04:37 have what they need to deploy these tools to production.
0:04:41 And you’re doing it kind of mentioned in the intro.
0:04:45 Trying to make the platform available for folks who are deep into this,
0:04:48 have been doing CV and working with machine learning for a while.
0:04:51 And then also folks who might be new to this, they can get up and
0:04:55 running and work with CV, build that into their toolkit.
0:05:00 >> Yeah, the emphasis has always been kind of on someone that wants to be a builder.
0:05:04 That definition is expanding with the capabilities of code generation,
0:05:05 prompting to apps.
0:05:08 We’ve always kind of been bent on this idea that those that
0:05:11 want to create, use, and distribute software.
0:05:14 What’s funny is that when we very first launched some of our first products,
0:05:16 ML teams initially were kind of like, I don’t know,
0:05:18 this seems pretty pedestrian, I know exactly what to do.
0:05:21 And fast forward now, and it’s like, whoa, a platform that’s fully
0:05:25 featured that has immediate access to the latest models to use on my data in
0:05:28 the contexts of where I can even anticipate.
0:05:31 So it’s been kind of, as the platforms become more feature rich,
0:05:35 we’ve been able to certainly enable a broader swath of both maybe
0:05:37 domain experts of a capability.
0:05:41 But I think broadly speaking, the places that we see the rapid,
0:05:44 most impactful adoption in some ways is actually bringing vision to others that
0:05:46 otherwise may not have had it.
0:05:51 Like what used to be maybe like a multi-quarter PhD thesis level investment
0:05:54 now can be something that a team spends up in an afternoon.
0:05:58 And that really has this kind of demand-begets-demand paradigm.
0:06:01 I mean, for example, one of our customers, they produce electric vehicles.
0:06:05 And when you produce an EV inside their general assembly facility,
0:06:08 there’s all sorts of things that you need to make sure you do correctly as you
0:06:12 produce that vehicle, from the worker safety who are doing the work to
0:06:16 the machines that are outputting, say, when you do what’s called stamping,
0:06:19 where you take a piece of steel or aluminum and you press it into the shape
0:06:23 of the outline of the vehicle, and you get up potential tears or fissures or
0:06:26 when you actually assemble the batteries out of the correct number of screws.
0:06:30 Basically, every part of building a car is about visually
0:06:33 validating that the thing has been built correctly so that when a customer
0:06:36 drives it, they can do so without any cause for pause.
0:06:40 And just a little bit of computer vision goes a really long way in enabling
0:06:44 that company and many others to very quickly accelerate their goals.
0:06:47 In fact, this company had the goal of producing 1,000 vehicles three years ago,
0:06:50 and they barely did it at 1,012 in that year.
0:06:53 And then they scaled up to 25,000 and now 50,000.
0:06:56 And a lot of that is on the backs of having things that they know they’re
0:06:57 building correctly.
0:07:01 And so we’re really fortunate to be a part of enabling things like this.
0:07:04 So it’s kind of like you could think about it of a system that doesn’t have
0:07:07 a sense of visual understanding, adding even just a little bit of visual
0:07:11 context totally rewrites the way by which you manufacture a car.
0:07:14 And that same revolution is going to take place for
0:07:16 lots of operationally intensive processes, but
0:07:19 any kind of place that you interact with the world each day.
0:07:23 >> So kind of along those lines, what to use the phrase untapped opportunities
0:07:27 you see out there, what’s the low hanging fruit or maybe the high hanging fruit,
0:07:30 but you’re just excited about when it comes to developing and
0:07:33 deploying computer vision applications.
0:07:37 And we’ve been talking about it, but obviously talk about robo flows.
0:07:40 Work not just supporting, but helping developers kind of unlike helping
0:07:42 builders unlock what’s next.
0:07:46 >> The amazing thing is actually the expanse of the creativity of developers
0:07:49 and engineers, it’s like if you give someone a new capability,
0:07:53 you almost can’t anticipate all the ways by which they’ll bring that capability
0:07:53 to bear.
0:07:57 So for example, I mean, we have folks that hobbyists that’ll make things,
0:08:00 like the number of folks that make things that like measure the size of fish.
0:08:03 Cuz I think they’re trying to prove to their friends that they caught the biggest
0:08:06 fish, and then you separately have like government agencies that have wanted to
0:08:09 validate the size of salmon during migration patterns.
0:08:12 And so this primitive of like understanding size of fish both has
0:08:14 what seems to be fun and very serious implications.
0:08:17 Or folks that I don’t know, like a friend of mine recently was like,
0:08:21 hey, I wonder how many cars out in San Francisco are actually Waymo’s,
0:08:22 versus like other sorts of cars.
0:08:23 And what does that look like?
0:08:24 What does that track over time?
0:08:28 And so they had a pretty simple like Raspberry Pi camera parked down their
0:08:31 windowsill and in an afternoon, now they have a thing that’s counting,
0:08:34 tracking, and keeping a tabulation on how many self-driving cars are making
0:08:37 their way on the road, at least sampled in front of their house each day.
0:08:40 >> Right, no, I don’t wanna call anybody out, but that’s not the same person
0:08:45 who had the video of all the Waymo’s in the parking lot in the middle of the
0:08:46 night in San Francisco in circles.
0:08:47 No, okay.
0:08:49 >> It wasn’t that guy.
0:08:53 >> Yeah, but I mean, like the use cases are expansive because I don’t know,
0:08:57 like the way we got into this, right, is like we were making AR apps, actually.
0:09:00 And computer vision was critical to the augmented reality,
0:09:02 understanding the system around us.
0:09:04 And we’ve since had folks that make, you know,
0:09:08 board game understanding, technology, D&D dice counters,
0:09:10 telling you the best first movie you should play in Catan.
0:09:12 And so basically, like you have this like creative population of folks.
0:09:15 Or this one guy during the pandemic, you know, he’s really bored, locked inside.
0:09:19 And he thought maybe his cat needed to get some more exercise.
0:09:22 And he created this system that like attached a laser pointer to a robotic arm.
0:09:26 And with a little bit of vision, he made it so the robotic arm consistently points
0:09:28 the laser pointer 10 feet away from the cat.
0:09:31 That’s like jumping around the living room and makes this whole YouTube tutorial.
0:09:33 But then like the thing that’s really interesting, right, is that like,
0:09:37 you know what technology has arrived when like a hacker can just like build
0:09:40 something in a single setting or maybe like in a weekend.
0:09:46 You know that what used to be this far difficult to access capability is now
0:09:50 broadly accessible. And that fuels like a lot of like the similar sort of enterprise
0:09:53 use cases. Like we got to have a joke at Rebelflow that like one person’s
0:09:56 hobbyist projects is another person’s entire business.
0:09:59 So the low hanging fruit, frankly, is like everywhere around us.
0:10:02 Like any sort of visual feed is untapped.
0:10:05 This sort of images that someone might be collection and gathering.
0:10:07 I mean, the similar things that give rise to machine learning certainly apply
0:10:11 to vision where the amount of visual inputs doubling year on year and petabytes
0:10:14 of visual information to be extracted.
0:10:16 So it’s kind of like, if you think about it, you can do it.
0:10:21 That makes me think of an episode we did recently with a surgeon who founded
0:10:26 a surgical data collective and they were using just these, I say stacks.
0:10:28 They weren’t actually videotapes, I’m sure.
0:10:33 But all of this unwatched unused footage from surgeries to train a model
0:10:36 to help train surgeons how to do their jobs better.
0:10:40 But if you want to ask, are the inputs that could, the visual inputs that can go
0:10:43 into RoboFlow, it doesn’t have to be a live camera stream.
0:10:46 You can also use, you know, archive footage images.
0:10:47 That’s correct.
0:10:47 Yep.
0:10:48 Yep.
0:10:51 So someone may have like a backlog of a bunch of videos.
0:10:54 I mean, for example, we actually had a professional baseball team where
0:10:58 they had a history of a bunch of their videos of pitching sessions and they
0:11:03 wanted to run through and do a key point model to identify various poses and time
0:11:06 of release of the pitch and how does that impact someone’s performance over time.
0:11:09 And so they had all these videos that from the past that they wanted to run through.
0:11:12 And then pretty soon they started to do this for like minor leagues where you
0:11:15 might not have scouts, omnipresent, and you certainly don’t have broadcasts.
0:11:16 Right, right, right.
0:11:19 You just have like this like kind of low quality footage on like cameras from
0:11:22 various places and being able to produce like sports analytics out of, you know,
0:11:26 this information that’s just locked up otherwise and this unstructured visual
0:11:30 capture is now available for, in this case, building a better baseball team.
0:11:31 Yeah, that’s amazing.
0:11:36 Building community, you know, is something that is both vital to a lot
0:11:39 of companies like tech companies and developer platforms and such.
0:11:43 But it also can be a really hard thing to do, especially to build, you know,
0:11:48 an organic, genuine, robust community serving enterprise clients, let alone
0:11:54 across this seemingly endless sort of swath of industries and use cases and such.
0:11:57 You know, also pretty resource intensive.
0:11:58 So how do you balance both?
0:12:02 How is Roboflow approaching, you know, building that community that you’re
0:12:08 talking about just now with serving, you know, these I’m sure demanding in a good
0:12:10 way, but, you know, demanding enterprise clients.
0:12:13 I think the two actually go hand in hand more than many would anticipate.
0:12:14 Okay.
0:12:18 When you build a community and you build a large set of people that are
0:12:23 interested in creating and using a given platform, you actually give a company
0:12:26 leverage basically like the number of people that are building, creating and
0:12:30 sharing examples of Roboflow from a very early day made us seem probably much
0:12:33 bigger than maybe we were or are.
0:12:36 And so that gives a lot of trust to enterprises.
0:12:40 Like, you know, you want to use something that has gone through its paces and
0:12:43 battle tested something that might be like an industry standard.
0:12:47 And you don’t become an industry standard by only limiting your technology to a
0:12:49 very small swath of people.
0:12:53 You enable anyone to kind of build, learn the paradigm and create.
0:12:58 Now, you’re right that both take a different type of thoughtfulness to be
0:12:59 able to execute on.
0:13:03 So in the context of like community building and making products for developers.
0:13:06 A lot of that I think stems from, you know, as an engineer, there’s
0:13:09 products that I like using in the ways that I like to use those products.
0:13:12 And I want to enable others to be able to have a similar experience of the
0:13:12 products that we make.
0:13:16 So it’s, it’s things like providing value before asking for value, having a very
0:13:19 generous free chair, having the ability to highlight the top use cases.
0:13:22 I mean, we have a whole like research plan where if someone’s doing stuff on a
0:13:26 .edu domain, then they have increased access to GPUs.
0:13:30 Roboflow has actually given away over a million dollars of compute and GPU usage
0:13:33 for open source computer vision projects.
0:13:37 And, you know, we actually have this, it’s kind of a funny stat, but 2.1
0:13:40 research papers are published every day, citing Roboflow.
0:13:43 And those are things like people are doing all these sorts of things.
0:13:44 That’s a super cool stat, I think.
0:13:48 Yeah, I mean, it just gives it, it just gives you the context to like, yeah,
0:13:52 like that is someone’s maybe six or 12 month thesis that they’ve spent trying
0:13:57 to inject and to be able to empower folks to realize what’s possible.
0:14:01 And it’s really kind of the fulfillment of like our mission at its core, like
0:14:05 the impact of visual understanding is bigger than anyone company and anything
0:14:09 that we can do to allow the world to see, expose and deploy that is important.
0:14:12 Now, on the enterprise side, what we were really talking about is building
0:14:15 a successful kind of go to market motion and making money to invest
0:14:19 further in our mission and enterprises, as you alluded, are very resource
0:14:23 intensive in terms of being able to service those needs successfully.
0:14:26 Even there, though, you actually get leveraged by seeing the sorts of
0:14:30 problems, seeing the sorts of fundamental building blocks and then productizing
0:14:33 those building blocks, you know, there have been companies that have come
0:14:37 before Roboflow who have done a great job of be very hands on with enterprise
0:14:40 customers and productizing those capabilities, company like Pivotal or
0:14:45 Palantir or these large companies that have gone from, hey, let’s kind of do
0:14:48 like a bespoke way of making something possible and deploy it more broadly.
0:14:52 Now, we’re not fully, you know, like for like with those businesses.
0:14:56 I more give that as an example to show as someone that is building tooling
0:15:00 and capabilities, worst case is you’re giving the enterprise substantially
0:15:03 more leverage and certainly best case is there’s actually a symbiotic
0:15:06 relationship between enterprises being able to discover how to use the
0:15:10 technology, be able to find guides from the community, be able to find models
0:15:11 they want to start from.
0:15:15 I mean, Roboflow Universe, which is the open source collection of data sets
0:15:18 and models, is the largest collection of computer vision projects on the web.
0:15:22 There’s about 500 million user labeled and shared images and over 200,000
0:15:23 pre-trained models.
0:15:26 And that’s used for the community just as much as enterprise, right?
0:15:28 Like when you say enterprise, like enterprise is people.
0:15:31 And so there’s people inside those companies that are creating and building
0:15:32 some of those capabilities.
0:15:35 Now, operationalizing and ensuring that we deliver the service quality, it’s
0:15:39 just the types of teams you build and the way that you prioritize companies
0:15:39 to be successful.
0:15:42 But we’re really fortunate that, you know, we’re not writing the playbook here.
0:15:47 There’s been a lot of companies that, you know, Mongo or Elastic or Twilio or
0:15:51 lots of post IPO businesses that have shown the pathway to both building
0:15:55 really high quality products that developers and builders love to use
0:15:59 and ensuring that they’re enterprise ready and meeting the needs of high
0:16:02 scale, high complexity, high value use cases.
0:16:04 So you use the word complexities.
0:16:09 And, you know, one of the things that I hear all the time from, and I’m
0:16:12 sure you more than me from people are trying to build anything is sort of how
0:16:17 do you balance, you know, creativity and coming up with ways to solve problems.
0:16:20 And particularly if you get into kind of a unique situation and you need to find
0:16:24 a creative answer with things, with not letting things get too complex.
0:16:27 And, you know, in something like computer vision, I’m sure the technical
0:16:30 complexities can, can spin up in a hurry.
0:16:33 What’s been, you know, your approach and what success is?
0:16:35 How have you found success and balancing that?
0:16:40 Complexity for a product or global flow is always a balance.
0:16:44 You want to offer users the capability in the advanced settings and the ability
0:16:46 to make things their own.
0:16:50 Well, also part of the core value proposition is simplifying.
0:16:53 And so you often think, oh man, those two things must be at odds.
0:16:55 How do you simplify something, but also serve complexity?
0:17:00 And in fact, they’re not, especially for products like Roboflow, where it is
0:17:03 four builders, you make it very easy to extend.
0:17:07 You make it very interoperable, meaning there’s open APIs and open SDKs, where
0:17:11 if there’s a certain part of the tool chain that you want to integrate with or
0:17:13 there’s a certain specific enterprise system where you want to read or write
0:17:16 data to, that’s all supported on day one.
0:17:19 And so if you try to kind of boil the ocean of being everything in the
0:17:24 platform all at once on day one, then you can find yourself in a spot where you
0:17:28 may not be able to service the needs of your customers well.
0:17:30 In fact, it’s a bit more step by step.
0:17:33 And that’s where, you know, the devil’s in the details of execution of which
0:17:36 steps you pick first, which sort of problems you best nail for your customers.
0:17:39 But philosophically, it’s really important to us that, for example, when
0:17:43 someone is building a workflow, which is, you know, the combination of a model
0:17:47 and some inputs and outputs, you might ingest maybe like an RTSP stream from
0:17:49 like a live feed of a camera.
0:17:51 Then you might have like a first model that’s, once you’re the problem that
0:17:57 we’re solving is we’re an inventory company and we’re concerned about worker safety.
0:18:00 You might have a first model that’s just like constantly watching all frames to
0:18:03 see if there’s a presence of a person, a very lightweight model kind of run in
0:18:04 the edge.
0:18:07 And then maybe a second model of when there’s a person, you ask a large vision
0:18:10 language model, large VLM, Hey, is there any risks here?
0:18:11 Is there anything to consider?
0:18:13 Should we like look more closely at this?
0:18:17 And then after the VLM, you might have a another specific model that’s going to
0:18:22 do validation of the type of danger that is interesting, or maybe the specific
0:18:25 area within your, your store, maybe you’re going to connect to another
0:18:26 database that exists.
0:18:29 And then like, based on that, you’re going to write some results somewhere.
0:18:32 And maybe you’re going to write that result to have insights of how frequently
0:18:37 there was a cause for concern within the process that you’re monitoring, just as
0:18:40 much as maybe you’re flagging an alert and maybe send in a text or an email or
0:18:43 writing to an enterprise system like SAP to keep track.
0:18:50 And at each step of that juncture, any one of those nodes, since it’s built for
0:18:54 us on an open source platform, which we call inference, you can actually mix
0:18:58 and match, write your own custom Python, write an API in way or one way or another.
0:19:02 And so let’s imagine like a future where someone wanted the ability to write
0:19:04 to a system that we didn’t support yet, like first party.
0:19:05 You’re actually not out of luck.
0:19:09 As long as that system accepts a post request, you’re fine.
0:19:11 And so you have the ability to extend the system.
0:19:15 Yeah. And so it’s this sort of paradigm of like interoperability and making
0:19:17 it easy to use with alongside other tools.
0:19:20 And it gets back to your point around servicing builders, just as much as the
0:19:25 enterprise, I actually think those things are really closely interlinked because
0:19:29 you provide the flexibility and choice and ability to make something mine and
0:19:33 build a competency inside the company of what it is I wanted to create and deploy.
0:19:37 Right. The way you frame that makes a lot of sense and makes that link very clear.
0:19:40 We’re speaking with Joseph Nelson.
0:19:43 Joseph is the co-founder and CEO of Roboflow.
0:19:48 And as he’s been detailing Roboflow provides a platform for builders to use
0:19:50 computer vision in what they’re building.
0:19:55 Joseph, you know, I in the intro sort of alluded to all the all the advances
0:19:58 and buzz around large language models and that kind of thing over the past couple
0:20:02 of years. And I meant to ask Roboflow was founded in 2020.
0:20:05 Roboflow Inc. was incorporated in 2020.
0:20:06 That’s right. Got it.
0:20:09 And so anyway, kind of fast forwarding to, you know, more recently, the past, I
0:20:15 don’t know, six months, year, whatever it’s been, a lot of buzz around agents,
0:20:17 the idea of agentic AI.
0:20:21 And then, you know, there was buzz, I guess, that the word multimodal was being
0:20:26 flung around kind of more frequently, at least in circles I run in for a while.
0:20:29 And then it sort of dropped off just as people, you know, there were the
0:20:33 consumer models that the Clause and ChatGPTs and Geminis and what have you
0:20:37 in the world, just started incorporating visual capabilities, both to, you know,
0:20:43 ingest and understand and then to create visual output, voice models, you know,
0:20:45 now getting into short video clips, all that kind of stuff.
0:20:51 What’s your take on the role of multimodal AI integration when it comes to advancing
0:20:55 CV, you know, how is Roboflow kind of positioned to support this?
0:21:02 So multimodality allows an AI system to have even more context than from a
0:21:03 single modality, right?
0:21:07 So if one of our customers is monitoring an industrial process, and let’s say
0:21:12 they’re looking for potentially a leak, maybe in an oil and gas facility, that
0:21:16 leak can manifest itself as, yes, you see something, a product that’s dripping
0:21:17 out and you didn’t expect it to.
0:21:23 It also can manifest itself as you heard a noise or maybe there’s something
0:21:27 about the time dimension of the video that you’re watching as another modality
0:21:28 beyond just the individual images.
0:21:33 Right. And those additional dimensions of data allow the system that you’re
0:21:35 building to have more intelligence.
0:21:39 And so that’s why you see kind of like all these modalities crashing together.
0:21:42 And what it does is it enables our customers to have even more context.
0:21:47 The way we’ve thought about that is we’ve actually been built on and using
0:21:50 multimodality as early as 2021.
0:21:54 So in 2021, there was a model that came out from open AI called the clip,
0:21:57 contrastive language image per training, which introduced this idea of training
0:21:59 on 400 million image text pairs.
0:22:03 Can we just associate some words of text with some images?
0:22:06 What this really unlocked for our customers was the ability to do semantic
0:22:10 search, like I could just describe a concept and then I can get back the images
0:22:11 from a video frame or from a given image.
0:22:15 Now it’d be interesting for me for the purposes of building out my model.
0:22:20 Increasingly, we’ve been excited by increases of models that have more
0:22:22 multimodal capabilities on day one.
0:22:26 That comes with its own form of challenges, though, the data preparation,
0:22:30 the evaluation systems, the incorporation of those systems into the other parts
0:22:32 of the pipeline that you’re building.
0:22:36 And so where there’s opportunity to have even more intelligence, there’s also
0:22:41 challenge to incorporating that intelligence, adapting it to your context,
0:22:42 passing it to other sorts of systems.
0:22:47 And so Roboflow and being deep believers in multimodal capabilities very early
0:22:53 on have continued to make it so that users can capture, use and process other
0:22:54 modalities of data.
0:22:58 So for example, we support the ability for folks to use vision language models,
0:23:01 BLMs, in the context of problems they’re working, which is typically like
0:23:02 an image text pair.
0:23:08 So if you’re using, you know, Quen VL 2.5, which came out last week, or
0:23:11 Florence 2 for Microsoft, which came out maybe about six months ago, or PolyGemma
0:23:16 2 from Google, these are all multimodal models that have very rich text
0:23:20 understandings and have visual understandings, which makes them very good at,
0:23:22 for example, document understanding.
0:23:26 Like if you just pass a document, there’s both text in the document and a
0:23:27 position in the document.
0:23:30 And so Roboflow is one of the only places, maybe the only place where you can
0:23:35 fine tune and adapt, say, Quen VL today, which means preparing the data and running
0:23:37 it in the context of the rest of your systems.
0:23:40 And those sorts of capabilities, I think, should only increase and enable our
0:23:44 customers to get more context more quickly from the types of problems that
0:23:44 they’re solving.
0:23:44 Right.
0:23:48 So I think a lot of these things kind of like are crashing together into just
0:23:52 like AI, like amorphous AI that like has all these capabilities, like you’d expect
0:23:57 it, but as that happens, what’s important is there’s actually still unique parts
0:24:00 of visual needs, right?
0:24:03 Like visual needs require visual tooling in our opinion.
0:24:05 Like you want to see, you want to validate.
0:24:09 You need to do, you know, the famous adage of a picture being worth a thousand
0:24:11 words is extremely instructive here.
0:24:14 Like you almost can’t anticipate all the ways that the world’s going to look
0:24:17 different than how you drew it up.
0:24:21 Like self-driving cars are kind of this example one-on-one where, yeah, you think
0:24:24 you can drive, like you have a very simple way of describing what the world’s
0:24:25 going to look like, but I don’t know.
0:24:29 Like let’s take a very narrow part of a self-driving car, stop signs, right?
0:24:31 So go stop signs look universal.
0:24:32 They’re always octagons.
0:24:35 They’re red and they’re really well-mounted on the right side of streets.
0:24:39 Well, what about a school bus where the stop sign kind of flips off where it
0:24:43 comes on or what about like a gate of where like the stop signs mounted on a
0:24:44 gate and the gate could open and close?
0:24:48 And pretty soon you’re like, wait a second, there’s a lot of cases where a
0:24:50 stop sign isn’t really just a stop sign.
0:24:55 And seeing those cases and triaging and debugging and validating, we think
0:24:59 inherently calls for some specific needs for processing the visual information.
0:25:04 And so we’re laser focused on enabling our customers to benefit from as many
0:25:08 modalities as help them solve their problem while ensuring the visual
0:25:11 dimension in particular is best capitalized on.
0:25:12 Right.
0:25:17 And does that, and I may be showing the limits of my technical understanding here.
0:25:19 So, you know, have added if so.
0:25:25 But does that exist as, you know, RoboFlow creating these, you know, sort of, as
0:25:30 you said, amorphous AI all crash together models that have this focus and these
0:25:31 sort of advanced visual capabilities?
0:25:38 Or is it more of a like chaining a RoboFlow specific model, you know, onto other models?
0:25:43 Commonly you’re in a position where you’re chaining things together or you
0:25:46 wanted things to work in your context or you wanted to work in a compute
0:25:47 constrained environment.
0:25:51 Okay, so visions, visions pretty unique in that unlike language and a lot of
0:25:55 other places where AI exists, actually vision is almost where humans are not.
0:25:58 Basically, like you want to observe parts of the world where a person is present.
0:26:02 Like if you return to our example of like an oil and gas facility where
0:26:06 you’re monitoring pipelines, I mean, there’s tens of thousands of miles of pipeline
0:26:09 and you’re certainly not going to have a person stationed every hundred yards
0:26:11 along it’s just an S9 idea.
0:26:15 And so instead you could have a video theater visual understanding of maybe key
0:26:20 points where you’re most likely to have pressure changes and to monitor those
0:26:23 key points, you know, that you’re not necessarily in an internet connected
0:26:27 environment, you’re in an operationally intensive environment that even if you
0:26:30 did have internet and might not make sense to stream the video to the cloud.
0:26:33 So basically where you get to is you’re probably running something at the edge
0:26:36 because it makes sense to co-locate your compute and that’s where like a lot of
0:26:39 our customers, for example, using video Jetsons, they’re very excited about the
0:26:44 digits that was announced at CES to make it so that you can bring these highly
0:26:48 capable models to co-locate alongside where their problem kind of exists.
0:26:50 Now, why does that matter?
0:26:54 That matters because you can’t always have the largest, most general model
0:26:56 running in those environments at real time.
0:27:00 I think this is part of, you know, a statement of like the way the world
0:27:04 looks today versus how we’ll look at 24, 36 and 48 months.
0:27:07 But I do think that over time, even as model capabilities advance and you can
0:27:10 get more and more distilled at the edge, there’s I think always going to be
0:27:14 somewhat of a lag between if I’m operating in an environment where I’m
0:27:17 fully compute unbounded or these comparatively unbounded in the cloud
0:27:20 versus an environment where I am a bit more compute bounded.
0:27:25 And so that capability gap requires specialization and capability to work best
0:27:27 for that domain context problem.
0:27:31 So a lot of Roboflow users and a lot of customers and a lot of deployments tend
0:27:34 to be in environments like those, not all, but certainly some.
0:27:39 All right, shift gears here for a moment before we wrap up.
0:27:41 Joseph, you’re a multi-time founder, correct?
0:27:45 Yeah, maybe to kind of set this up, you can just kind of run through a little
0:27:48 bit your experience as an entrepreneur.
0:27:49 What was the first company you founded?
0:27:53 Well, the very first company was a T-shirt business in high school.
0:27:56 Nice. I don’t know that it was founded, there’s never an LLC.
0:27:58 I don’t even know my parents knew about it.
0:28:02 But there is that in a university.
0:28:08 I ran a satirical newspaper and sold ads on the ad space for it and date myself here.
0:28:11 But Uber was just rolling out to campuses at that time.
0:28:14 So I had my Uber referral code and had like free Ubers for a year for like all
0:28:16 the number of folks that discovered it.
0:28:20 I kind of joked my first company that maybe the closest thing to a real business
0:28:24 beyond these side projects was a business that I started my last year of university
0:28:27 and ran for three years before a larger company acquired it.
0:28:29 And I went to school in Washington, D.C.
0:28:34 I had interned on Capitol Hill once upon a time and I was working at Facebook
0:28:38 my last year of university and was brought back to Capitol Hill and realized
0:28:41 that like a lot of the technical problem or a lot of the problems,
0:28:44 operational problems that could be solved with technology still existed.
0:28:48 One of those is Congress gets 80 million messages a year and interns sort through
0:28:51 that mail. And this was, you know, 2015.
0:28:55 So we said, Hey, what if we use natural language processing to accelerate
0:28:57 the rate at which Congress hears from its constituents?
0:29:01 And in doing so, we improve the world’s most powerful democracies,
0:29:02 customer success center.
0:29:07 And so that grew into business that I ran for about three years and we had a tight
0:29:10 integration with another product that was a CRM for these congressional offices
0:29:13 and that company called Fireside 21 acquired the business and rolled it out
0:29:15 to all of their their customers.
0:29:19 That was a bootstrap company, you know, as nine employees at PE can relatively
0:29:23 mission driven thing that we wanted to build and solve a problem that we knew
0:29:26 should be solved, which is improving the efficacy of Congress.
0:29:28 How big is Roboflow?
0:29:28 How many employees?
0:29:31 Well, I tell the team, whenever I answer that question, I start with,
0:29:33 we’ve helped a million developers so far.
0:29:37 So that’s how that’s how big we are team wise, team wise.
0:29:40 Team doesn’t necessarily mean, you know, come in any, any number of things.
0:29:43 Yeah, yeah, we’re growing quickly.
0:29:43 Excellent.
0:29:49 As we’re recording this and this one’s going to get out before GTC 2025 coming
0:29:52 up in mid-March down in San Jose, as always.
0:29:54 And Joseph Roboflow is going to be there.
0:29:55 Yeah, we’ll be there.
0:29:56 I mean, GTC has become the Super Bowl of AI.
0:29:57 Right.
0:30:01 Any hints, any teasers you can give of what you’ll be showing off?
0:30:05 We have a few announcements of some things that we’ll be releasing.
0:30:08 I can give listeners a sneak peek to a couple of them.
0:30:13 One thing that we’ve been working pretty heavily on is the ability to chain models
0:30:16 together, understand their outputs, connect to other systems.
0:30:21 And from following our customers, it turns out what we kind of built is a system
0:30:26 for building visual agents and increasingly as there’s a strong drive around
0:30:29 agentic systems, which is, you know, more than just a model.
0:30:33 It’s also memory and action and tool use and loops.
0:30:38 Users can now create and build and deploy visual agents to monitor a camera feed
0:30:42 or process a bunch of images or make sense of any sort of visual input in a
0:30:45 very streamlined, straightforward way using our open source tooling in a
0:30:46 loginless way.
0:30:50 And so that’s one area that we’re excited to show more about soon.
0:30:55 In partnership with NVIDIA and the inception program, we’re actually
0:30:59 releasing a couple of new advancements in the research field.
0:31:03 So without giving exactly what those are, I’ll give you some parameters of what
0:31:09 to expect at CBPR in 2023, Robofo released something called RF 100, which
0:31:12 the premise is for computer vision to realize its full potential, the models
0:31:15 need to be able to understand novel environments.
0:31:15 Right.
0:31:18 So if you think about a scene, you think about maybe people on a restaurant
0:31:21 or you think about like a given football game or something like this.
0:31:22 Yeah, yeah.
0:31:24 But the world is much bigger than just where people are.
0:31:25 Like you have like documents to understand.
0:31:27 You have aerial images.
0:31:28 You have things under microscopes.
0:31:30 You have agricultural problems.
0:31:31 You have galaxies.
0:31:36 You have digital environments and RF 100, which we released is sampling
0:31:40 from the Robofo universe, a basket of a hundred data sets that allows
0:31:45 researchers to benchmark how well does my model do in novel contexts.
0:31:47 And so we really sat in 23.
0:31:52 And since then labs like Facebook, Apple, Baidu, Microsoft, NVIDIA, Omniverse
0:31:55 team have benchmarked on what is possible.
0:31:59 Now, universe has, the Robofo universe has grown precipitously since then as
0:32:02 I have the types of challenges that people are trying to solve with computer
0:32:07 vision. And so we’re ready to show what the next evolution of advancing
0:32:11 visual understanding and benchmarking understanding might look like at GTC.
0:32:16 And then a second thing we’ve been thinking a lot about is the advent of
0:32:21 transformers and the ability for models to have really rich pre trainings
0:32:24 allows you to kind of start at the end, so to speak, with a model and its
0:32:31 understanding, but that hasn’t fully made its way as impactful as it can to vision,
0:32:35 meaning like how can you use a lot of the pre trained capabilities and especially
0:32:37 to vision models running on the edge.
0:32:41 And so we’ve been pretty excited about how do you marry the benefits of
0:32:45 pre trained models, which allow you to generalize better with the benefits of
0:32:46 running things real time.
0:32:51 And so actually this is where NVIDIA and Robofo have been able to pair up
0:32:55 pretty closely on something that we’ll introduce and I’ll leave it at that for
0:32:57 folks to see and do an interesting to learn more.
0:33:00 All right, I’m signed up.
0:33:01 I’m interested, can’t wait.
0:33:05 So you’ve done this a few times and you know, one way or another, I’m sure
0:33:09 you’ll do it again going forward and you know, scaled up and all that good stuff.
0:33:14 Lessons learned advice you can share, you know, for founders, for people out
0:33:18 there thinking about and you know, whether it’s CV related or not.
0:33:19 What does it take?
0:33:22 What goes into being, you know, being a good leader, building a business,
0:33:27 taking an idea, seeing it through to a product that, you know, serves humans
0:33:29 as well as solving a problem.
0:33:32 What wisdom can you drop here on listeners thinking about their own
0:33:34 entrepreneurial pursuits?
0:33:36 One thing that I’ll know is you said you’ll do it again.
0:33:40 I’m actually very vocal about the fact that Robofo is the last company
0:33:44 that I’ll ever need to start like the lifetime’s worth of work by itself.
0:33:48 As soon as I said it, I was like, I don’t know that.
0:33:49 He doesn’t know that.
0:33:50 And what if that comes off?
0:33:52 Like Roboflow is not going to, I was thinking about, oh, your last
0:33:55 company got acquired and so on and so forth, but that’s great.
0:33:59 I mean, that’s like in and of itself, you know, I suppose that could be
0:34:03 turned into something of a motto for aspiring entrepreneurs or what have you.
0:34:07 But that’s instructive actually for your question because I think a lot of people,
0:34:11 you know, you should think about the mission and the challenge that you’re,
0:34:13 you know, people say commonly like, oh, you’re marrying yourself to you
0:34:17 for 10 years, but I think even that is perhaps too short of a time horizon.
0:34:21 It’s what is something that you like a promise face that you can work on
0:34:25 excitedly in the world is different as a result of your efforts.
0:34:28 I will also note that, you know, what does it take?
0:34:29 How does it figure it out?
0:34:30 I’m still figuring it out myself.
0:34:32 There’s like new stuff to learn every single day.
0:34:37 And I can’t wait for like every two years when I look back and just sort
0:34:40 of cringe at the ways that I did things at that point in time.
0:34:44 But I think that, you know, the attributes that allow people to do well in startups,
0:34:49 whether they’re working in one, starting one, interacting with one is a deep sense
0:34:55 of grit and diligence and passion for the thing that you’re you’re working on.
0:35:00 Like the world doesn’t change by itself and it’s also quite malleable place.
0:35:05 And so having the wherewithal and the aptitude and the excitement and vigor
0:35:12 to shape the world the way by which one thinks is possible requires a lot of drive
0:35:16 and determination. And so, you know, it’s work with people, work in environments,
0:35:22 work on problems where if you have that problem changed with that team and the
0:35:27 result that that company that you’re working with continues to be realized.
0:35:28 What does that world look like?
0:35:29 Does that excite you?
0:35:33 And does it give you the ability to say independently, I would want to day in
0:35:37 and day out, give it my best to ensure and realize the full potential here.
0:35:41 And when you start to think about your time that way of something that is a
0:35:46 mission and important and time that you want to enjoy with the team, with the
0:35:50 customers, with the problems to be solved, the journey becomes the destination
0:35:51 in a lot of ways.
0:35:53 And so that allows you to play infinite games.
0:35:57 It allows you to just be really focused on the key things that matter and
0:36:01 delivering customer value and making products people love to use.
0:36:03 And so I think that’s fairly universal.
0:36:06 Now, in terms of specific advice, one things or another, there’s a funny
0:36:11 paradox of like advice needs to be adjusted to the prior of one situation.
0:36:14 It’s almost like the more universally useful the piece of advice is perhaps
0:36:17 like the less novel and insightful it might be.
0:36:17 Right.
0:36:21 Here I’ll note that I pretty regularly learn from those that are a few stages
0:36:23 ahead of me and aim to pay that favor forward.
0:36:27 So I’m always happy to be a resource for folks that are building or navigating
0:36:30 career decisions or thinking about what to work on and build next.
0:36:32 So I’m pretty findable online and welcome that from listeners.
0:36:33 Fantastic.
0:36:38 So let’s just go with that segue then for folks listening who want to learn
0:36:43 more about Roboflow, want to try Roboflow, want to hit you up for advice
0:36:45 on working at or with a startup.
0:36:47 Where should they go online?
0:36:50 Company sites, social medias, where can listeners go to learn more?
0:36:54 Roboflow.com is where you can sign up from the build of the platform.
0:36:55 We have a careers page.
0:36:59 If you’re generally interested in startups, work@astardup.com is YC’s
0:37:02 job support and we’ve hired a lot of folks from there.
0:37:03 So that’s a great resource.
0:37:10 I’m accessible online on Twitter of our ex @JosephofIowa and regularly share
0:37:11 a bit about what we’re working on.
0:37:13 And I’m very happy to be a resource.
0:37:16 If you’re in San Francisco and you’re listening to this, you might be surprised
0:37:18 that sometimes I’ll randomly tweet out when we’re welcoming folks to come
0:37:21 co-work out of our office on some Saturdays and Sundays.
0:37:23 So feel free to reach out.
0:37:24 Excellent.
0:37:25 Just Nelson, Roboflow.
0:37:27 This is a great conversation.
0:37:29 Thank you so much for taking the time.
0:37:33 And, you know, as you well articulated, the work that you and your teams are
0:37:39 doing is not only fascinating, but it applies to so much of what we do on
0:37:40 the earth, right, and beyond the earth.
0:37:45 So all the best of luck in everything that you and your growing community are doing.
0:37:46 Really appreciate it.
0:37:48 [MUSIC PLAYING]
0:37:50 [MUSIC PLAYING]
0:37:52 [MUSIC PLAYING]
0:37:54 [MUSIC PLAYING]
0:37:56 [MUSIC PLAYING]
0:37:58 [MUSIC PLAYING]
0:38:00 [MUSIC PLAYING]
0:38:02 [MUSIC PLAYING]
0:38:04 [MUSIC PLAYING]
0:38:06 [MUSIC PLAYING]
0:38:08 [MUSIC PLAYING]
0:38:10 [MUSIC PLAYING]
0:38:12 [MUSIC PLAYING]
0:38:14 [MUSIC PLAYING]
0:38:16 [MUSIC PLAYING]
0:38:18 [MUSIC PLAYING]
0:38:22 [MUSIC PLAYING]
0:38:24 [MUSIC PLAYING]
0:38:26 [MUSIC PLAYING]
0:38:28 [MUSIC PLAYING]
0:38:30 [MUSIC PLAYING]
0:38:32 [MUSIC PLAYING]
0:38:34 [MUSIC PLAYING]
0:38:36 [MUSIC PLAYING]
0:38:46 [BLANK_AUDIO]

0:00:10 [Music]
0:00:15 Hello and welcome to the NVIDIA AI podcast. I’m your host, Noah Kravitz.
0:00:19 In late 2024, the Telenoar AI factory was officially launched.
0:00:24 The AI factory is Norway’s first sustainable, sovereign, and secure
0:00:28 generative AI cloud service designed to enhance AI adoption for both internal
0:00:33 operations and external customers and to provide local AI computing
0:00:36 capabilities to the Nordic region. With us to share the story behind the
0:00:41 Telenoar AI factory and to discuss the impact responsible AI is set to have on
0:00:45 the country of Norway and the Nordic region more broadly is Karin Hilson.
0:00:50 Karin is the chief innovation officer and head of the AI factory at Telenoar.
0:00:55 And she’s also set to speak at NVIDIA GTC 2025 as part of a session titled
0:01:00 Accelerating Sovereign AI Factories – Insight from Telco Case Studies.
0:01:03 Karin, welcome to the AI podcast and thank you so much for taking the time to
0:01:07 join us. Thanks, Noah. Wow, what an introduction.
0:01:10 Well, you know, it’s hard because I try to, you know, I want to give you your
0:01:14 credit, give you your flowers as they say, but our guests and your no exception
0:01:17 have done so much. It’s hard to cram it all in there, but it’s all genuine.
0:01:21 We’re delighted to have you on the podcast. It’s great to be here, Nara.
0:01:26 So before we get into the story of the AI factory, which I’m really excited to hear
0:01:30 from you, maybe you can start by telling us a little bit about your own background
0:01:33 and your journey into AI. And then we can talk about Telenoar.
0:01:36 Yeah, I mean, the journey into AI, I mean, do we all end up here?
0:01:42 It’s a bit of the question, but is that my journey is maybe being shortly.
0:01:47 But I’ve been working in the the telco industry and with Telenoar now for 25 years
0:01:51 across the globe in many different continents and sort of moving back to
0:01:55 Norway a couple of years ago. And in the last few years, I’ve been working very
0:02:01 much with innovation, with sustainability and these things. And a bit of a fun fact,
0:02:07 I was actually sort of read up the other day that in, was it 2023 that
0:02:11 emissions that data centers have actually reached the same level as emissions
0:02:16 as the global airline industry. And that was a bit of an eye opener for me.
0:02:19 I don’t know if it’s fact or fiction, but it just sort of got me thinking.
0:02:24 And and then I see with the AI now sort of driving demand and everything,
0:02:27 the data sense is going to increase. And then I was thinking, well, okay,
0:02:32 but then we need to, how can we do this responsibly? And as I said, my journey to it,
0:02:36 I was working a lot with sustainability and how we can actually use our digital
0:02:44 infrastructure in the green shift. And then the vision of sort of democratizing AI
0:02:49 and every nation is an AI sort of came in. And I thought, wouldn’t it be cool if we
0:02:53 could build a green AI factory? And it was very much sort of the, the
0:02:57 passion I had around the sustainability, the need for AI is sort of,
0:03:02 it just beautifully came together. And I remember in the summer, sort of standing
0:03:06 on the banks of the Thames with my friends at a pub in the UK. And they said to me,
0:03:08 oh, come on, what are you doing at the moment? And I said, oh, yeah,
0:03:12 I’m building a green AI factory. And they said, is that even possible?
0:03:21 So I just love, so it’s sort of how I ended up, but I guess it’s a passion to make ideas happen.
0:03:26 Absolutely. You were working on sustainability previously in the telco industry?
0:03:30 Yeah, I’ve had various roles, but I had a sort of passion to sort of,
0:03:36 I was working with the critical infrastructure. I’ve had various CEO roles in Sweden,
0:03:41 in Montenegro, I’ve been working in Asia. And as I said, move back to Norway for
0:03:47 private reasons and looking, what can we really sort of do now? And seeing more and more that
0:03:54 the telcos critical infrastructure has such a key role in society. And I think this was one of my
0:04:00 eye knows during COVID, I think, like many of us sit and reflect. And I think all of us became so
0:04:06 dependent on the critical infrastructure that we have and the digital infrastructure.
0:04:10 So let’s talk about the factory. Should we start on the sustainability angle?
0:04:16 Should we start on sort of the commercial and sort of benefit of the factory to the
0:04:20 community? How would you like to begin? Tell us the story of building the AI factory?
0:04:25 I mean, I always like to sort of say, the story always started with a vision. And it was there
0:04:31 are sort of a top executive teleno where we’re meeting the sort of top executives in Nvidia.
0:04:35 I love this standard. There’s sort of every country must own the production of their own
0:04:42 intelligence. I sort of love that. And that sort of vision, then with the combination of teleno,
0:04:48 we’re 170 years old. We’ve not been running critical infrastructure in Norway for that time.
0:04:55 That’s amazing. It’s just such a beautiful fit. And I just love this vision that we had, okay,
0:05:00 that every country needs to sort of own his own intelligence and produce it. We need to do this
0:05:06 in a responsible, sustainable way. And we need critical infrastructure to do this. So this was
0:05:11 sort of the idea was then born. And I remember executives sort of coming back to Norway and
0:05:16 sort of said to me, current, can you make this happen? And he’s like, okay, do I have any funding?
0:05:22 Do I have no, no, just make it happen. And it was like, okay, where do we start sort of thing?
0:05:28 And every, you know, and everybody said, just believe, believe. And this is where our journey
0:05:34 started. There’s sort of which angle do you sort of tackle it? And if anything, I was overwhelmed.
0:05:39 And I think the only thing was certain is there was so much uncertainty, so many unknowns.
0:05:45 And we very much then started the journey exploring, okay, is there a need for it in
0:05:52 Norway? What’s the level of maturity? How can we build on this? And we started then working in
0:05:58 sprints. Forgive me for interrupting Karen, but when was this? How long ago? So this was, we started
0:06:02 the journey in February last less than a year ago. Less than a year ago. It’s amazing how fast
0:06:07 things are moving. Yeah, it was, even when I say, do you know, I feel like I’ve been working on a
0:06:15 lifetime, but it’s just just the sort of a year ago. And we started looking up the market, we
0:06:20 started looking, okay, what is our go to market proposition? You know, what equipment do we need
0:06:27 and what data center do we do it in? And all these sort of questions were coming up. And this is
0:06:34 quite early on we said, okay, you know, telcos are fantastic, but they maybe haven’t got the best
0:06:39 reputation for moving fast. So it’s like, how could we do it? And this is where we said, okay,
0:06:45 let’s take a bit of a different approach. And even the basic thing I faced with like, okay,
0:06:50 I want to order our first sort of cluster to put in a data center. And I didn’t have a legal entity
0:06:56 to do this with. I mean, a very silly sort of internal governance. So we decided, okay, let’s
0:07:02 think like a setting up like a startup, we’re going to set up a new company. So we set up a new
0:07:07 legal entity. And we said, okay, we’re going to be a startup. I said, I don’t want a huge team,
0:07:12 I don’t want a huge program, which is going to sort of drag us down in corporate governance,
0:07:18 slow decision making. So we said, okay, we’re going to think like a startup, we set up then our
0:07:26 first cluster, we caught together our go to market proposition. And then in August, we started to go
0:07:32 talk to some customers out in the market and said, this is our proposition, we’re going to have,
0:07:37 we’ve got an offering where we can offer a sovereign, secure and sustainable effectively.
0:07:43 What do you think? And this is sort of them where I think the fun really, really started.
0:07:49 And I always say, I said to the organization as well, the decision makers, the next talk I’m
0:07:55 going to come is when we’ve signed two customers. And I even remember standing on stage at an
0:08:00 internal town hall. And one of the leaders said to me, Karm, what does success look like? At the
0:08:05 end of the year, what does success look like? And I spontaneously said, yeah, we have two customers
0:08:11 in our AI factory. And there was, okay, that is Ray. And I said, yeah, we’re going to have two
0:08:16 customers, one internal and one external. And I was like, okay, so that became our goal. And this
0:08:22 is where the team then were fully focused that this is what we would then work towards.
0:08:24 Was that high autonomy, the first external customer?
0:08:31 So high, actually, we actually started off saying we wanted to tell another as our own
0:08:38 internal customer. Because we said it was very good to sort of start with that, to learn, build
0:08:43 confidence, you know, make sure we’re there before we dare go out and talk to other customers.
0:08:44 Right, right.
0:08:49 And this is where we pivoted quite early is then we need more intel if we just focus on one
0:08:54 customer. And again, we did a lot of this in building the AI factory for our own needs.
0:09:01 We saw this capability being fundamental for Teleno to accelerate his own AI journey.
0:09:07 Right, as well. I mean, Teleno sits on a lot of critical information and data,
0:09:13 which we need to secure that is sort of stays within in Norway and is operated very, very
0:09:19 securely. So we actually built the AI factory originally for our own needs. But we saw that
0:09:25 there was also a demand outside Teleno. So we pivoted quite quickly saying we can do this in
0:09:32 parallel, get one internal customer and one external customer. So we’re working sort of with
0:09:38 the Teleno operations to say, okay, they come on board. And then as you mentioned,
0:09:41 Hive Autonomy were then the actual first external customer.
0:09:41 Right, okay.
0:09:46 We got a board. And I think from both, you know, we’re learning a lot from both our customers.
0:09:52 And we have an MVP product. It’s not perfect. We’ve set some design principles that the whole
0:09:59 team, whereas where we say speedover perfection, we always like to say that roughly right is
0:10:05 better than precisely wrong. And again, in a Telco, we are so, for years, we’ve been on the
0:10:10 building, you know, perfect network. So we’re kind of our comfort zone here. And again,
0:10:16 in finding the right and what we call our MVP customers was quite critical. So we can have
0:10:19 customers that can learn and grow with us.
0:10:25 Are there currently specific use cases that you’re tackling, whether on the internal side,
0:10:29 or, you know, that Hive or perhaps other external customers are interested in? What are
0:10:31 some of the leading use cases?
0:10:35 Yeah. So I think that the one thing all of them have in common, first of all, is sort of very
0:10:40 giant. They’re all solving a specific problem. It’s not just like, oh yeah, we’ll buy some GPUs,
0:10:42 and then we’ll work out what we’re going to do.
0:10:45 Right, right, right. What can this chatbot do? Let’s play around with it.
0:10:50 Yeah, exactly. Which is what I think a lot of people want to do. But as I know, we want to
0:10:57 sort of, so like Hive autonomy, for example, I mean, they work with logistics, robotics.
0:11:03 So they are actually innovating, I say a lot of industries, whether it’s ports, as I said,
0:11:10 factories or this sort of in their operations and have efficiency cases. So they have very
0:11:16 specific customer needs that they are trying to solve. The reason sort of why they were
0:11:20 very interested in coming to the factory is that they’re sitting with sensitive data. So it was
0:11:27 very keen, they wanted it to be really on Norwegian soil. The teleno grand sort of represents
0:11:32 security, you know, so there’s sort of a gain that really helps them. And then the sustainability
0:11:39 part is super key. And so that was sort of the combination of these three. Cup Gemini is also
0:11:46 a customer of ours. They are developing products of doing voice to voice translation. And we can
0:11:51 say, yes, that can be done. But these are for sensitive dialogues. Not all dialogues can go
0:11:56 out in the cloud somewhere. These are very sort of sensitive dialogues, if you think, you know,
0:12:02 within the health sector, within the police. So not so much on prep, but again, it’s sort of a
0:12:10 safe, secure environment. And that’s really key. And another customer is working a lot with the
0:12:17 municipalities in Norway. And again, with sort of sensitive cases that they sort of really would
0:12:25 like their data to be secured. And the sustainability part of it. And this is something that, again,
0:12:30 as I said to you in my intro, I’m very, very passionate about. And then sort of, you know,
0:12:36 while some people say AI can solve climate change. So I think, you know, with the increasing
0:12:41 number of compute power and data centers that are needed, we have to be responsible and build
0:12:49 data centers in a sustainable way. And this is also why Telenoor is also building with several
0:12:56 partners, a state of the art modern data center here in Oslo. So Telenoor, in addition to the AI
0:13:02 factory is also partnered with a leading power company called Hufsman, and also a renewable
0:13:08 energy investor called high tech vision are actually now building a super modern data center.
0:13:14 And here it’s actually, this is not just about using renewable energy sort of coming in,
0:13:20 but is also all the excess heat will actually go into district heating and actually heat up
0:13:25 apartments and the surrounding area. So the heat coming out of the data center, you’re going to
0:13:30 capture it and redistribute it to heat homes in the area. Yes. I have very little, you know,
0:13:34 physical world building and engineering capabilities. So it’s a genuine question.
0:13:40 Is that a tricky undertaking? Or is it? Yes, it is. And this is sort of a company,
0:13:46 which is sort of Telenoor is partnering, taking all that. So I will say, we know. But I think what
0:13:53 we need to provide is then sort of, again, the sort of the critical infrastructure, the connectivity,
0:13:59 the security elements of it, and then being able to say to customers, and we’re doing this in an
0:14:05 energy efficient way. Our guest is Karin Hilson. Karin is the chief innovation officer and head
0:14:10 of the AI factory at Telenoor. And Karin Telenoor, we’ve talked Telenoor Norway,
0:14:14 Telenoor Sweden, and forgive me, I should have asked this up front. How many countries,
0:14:19 how big of an area does Telenoor serve? Telenoor, I mean, globally has over 200 million customers.
0:14:26 We have footprints in the Nordics. So Telenoor has a presence then in Norway, Sweden, Denmark,
0:14:32 and Finland. And then we have a presence in Asia. Got it. So to get back to the AI factory and thinking
0:14:38 about the use cases and such, are most of the use cases right now, and again, whether, you know,
0:14:44 actually happening or sort of in the works, are they centering around generative AI and large
0:14:51 language models? Or is it kind of, you know, other forms of machine learning and AI? What’s the buzz
0:14:56 right now in terms of, you know, the sort of current and future-looking use cases for the
0:15:02 factory? I would say that there is a lot of buzz that’s still in the exploratory phase to be very
0:15:07 open and honest. We forget. That’s where we all are. We’re still so early in all of this.
0:15:14 Yeah. And I see that there is definitely different maturity levels. I mean, when we talk to customers,
0:15:20 the market, and so many are, you know, exploring, you know, some very sort of super cool and
0:15:26 forward-leaning others are very much like, okay, again, going back to the, we’re sitting on very
0:15:33 critical data and everything. And it’s used to being in our, you know, basement or under our desks
0:15:38 and everything. But how can we sort of, we need more compute power and everything. And these are
0:15:44 the kind of dialogues that we are having with customers, sort of, how can we really, you know,
0:15:51 secure the handling of their data in a very secure way? They can trust that it is sort of,
0:15:58 stays on Norwegian soil that we still then are owning the production of intelligence as we
0:16:05 spoke about earlier. So I would say there is still, the dialogues are still very much evolving
0:16:08 around this. You’ve worked in different regions of the world and you currently do,
0:16:15 as you were just speaking to, how similar or different are not just the laws and regulations,
0:16:23 but sort of the common wisdom, the attitudes around responsible AI and data security, those
0:16:28 kinds of things. And these are obviously, you know, new and evolving topics, responsible AI
0:16:34 specifically. How similar or different do you find conversations about these things as you work in
0:16:39 different countries, different regions of the world? I would say they are very different.
0:16:46 I say the similarity is that certainly around security and sovereignty, what I have seen just
0:16:52 on the journey of the last two months is really increased. We are onboarding one customer at
0:17:00 the moment into the AIF factory and they sent us a list of 135 security questions. And this is sort
0:17:08 of what we forget. And again, these are becoming very, not just topical, but very sort of business
0:17:16 critical questions as well. And this is back to why we started the journey with the AI factory.
0:17:21 We saw that there was a bit of a hole in the market, if I could say that, we’ve been able
0:17:27 to have sovereignty, security and sustainability all three together. And then using the sort of
0:17:33 teleno trusted brand as when people trust us, we’re reliable, you know, is the sort of
0:17:39 combination. I mean, one could say, if you just want, you know, compute capacity to build an open
0:17:45 sort of model, go ahead and do that. But there is a lot of sort of really to move society forward
0:17:50 as a whole. You know, if it’s in the health sector, within defense, within the public sector, we see
0:17:56 that, you know, that there are sensitive dialogues, we see that there are, you know, things. So we
0:18:03 really see that the teleno AI factory, and together with our partnership with NVIDIA can
0:18:08 really help bring society forward. Absolutely. Because it’s some of these dialogues, as you say
0:18:14 now, around sovereignty and security, that are making people nervous, they’re making people
0:18:20 uncertain, they’re making people relook at their, you know, their fantastic IT, whatever that they’ve
0:18:26 had for the last years. So we’ve seen that there is definitely a change in climate, I would say,
0:18:33 around the seriousness of the dialogues. It seems appropriate, it makes sense to me.
0:18:38 All right, Karin, before we get to wrapping up, are there AI tools that you’re finding particularly
0:18:43 helpful, or, you know, just that you’re using regularly in your own life, be it work outside of
0:18:47 work, you don’t have to go into the details of what, but are there any tools that, you know,
0:18:52 you’ve been using regularly recently? Well, I have to confess, probably that I use it most to
0:18:57 help my kids with their homework, or if they say I’m cheating, I always tell them they can’t,
0:19:03 but I need to just answer all the, you know, I have to, you know, to answer all the why
0:19:09 questions and what it was. Yes, my older child tells me actually, when he’ll use it to sort of,
0:19:13 it’s almost the way that I use in people, we talk about on the show, people using, especially
0:19:18 folks in creative lines of work or doing creative, you know, projects that you’ll do something and
0:19:23 then send it to a chat bot to kind of get its take on it. And it kind of points out, oh, you missed
0:19:26 this, or, oh, here’s another way to think about that kind of thing. So I don’t think it’s cheating
0:19:31 myself, but, you know. No, it’s not cheating, but it helps get a dialogue because then exactly,
0:19:36 as you say, I then engage in discussions and we talk about whether it’s right or wrong,
0:19:41 and it is a fact that it actually, I find it actually is not cheating, I joke, but it actually
0:19:47 is a trigger to get the dialogue going. Which is fantastic, that’s great. Which isn’t always easy
0:19:53 with teenagers. No, no, it’s not. No, it’s not. All right, I want to give you time to talk about
0:19:58 the vision going forward. You know, this began with a vision, and it’s a vision that’s taken
0:20:03 flight now. What are your hopes? What’s your vision for the AI factory in the next, we’re
0:20:09 recording this in February 2025. So whatever time period makes sense, as we said, things are
0:20:13 moving so fast, it’s hard to know where AI will be in three years. But, you know, the next year,
0:20:17 two years, five years, what are you, where are you taking the AI factory?
0:20:21 Yeah, I mean, we certainly want to scale it. As I said, you know, we see, we’re starting in Norway,
0:20:30 we see Norway has this need. And I really believe that we can really help empower society here in
0:20:36 Norway through giving, you know, access to a place where different organizations, business,
0:20:42 it is the public sector, the private sector, can really innovate their businesses, make it more
0:20:49 efficient. And really sort of going back to this and feel that, yes, we are really doing this in a
0:20:56 sort of sustainable way. We’re helping that we’re really, we know our data and it’s safe,
0:21:01 we can run these really sensitive cases. And it’s sort of going back to this, I just love
0:21:07 this, the production of the intelligence is here in Norway. So really, I do see the AI factory as
0:21:12 being a very, maybe I’m being too, my vision is getting a bit, but the more I work with it,
0:21:17 the more I see it can really help move societies forward and develop.
0:21:21 Fantastic. That’s what it’s all about, if you ask me, but just my opinion.
0:21:23 My fully agree.
0:21:27 Karen Hilson, this was a pleasure. Thank you for taking the time. And you’ll be on the panel at
0:21:33 GTC next month as we record this. So looking forward to that. And for folks listening,
0:21:39 who would like to find out more about the AI factory, about Telanor, anything we’ve discussed?
0:21:44 Is there a good website, URL, social media, where can listeners go online to learn more?
0:21:49 Yeah. I mean, go onto the Telanor website, you’ll find our Telanor AI factory there. You can reach
0:21:57 out to me on LinkedIn as well. So please, we’re welcome. We’re on this journey together. As you
0:22:02 say, we’re still very much in an explorative phase. So I’d love to hear from people.
0:22:06 Fantastic. Well, again, Karen, thank you so much. It’s been a pleasure and
0:22:09 all the best of luck with all the work you’re doing. There’s no better reason to employ AI
0:22:12 than to bring us all forward, as you said. So all the best.
0:22:14 Thanks, Snare.
0:22:18 [Music]
0:22:29 [Music]
0:22:40 [Music]
0:23:01 [Music]
0:23:10 [BLANK_AUDIO]

0:00:10 [MUSIC]
0:00:13 Hello, and welcome to the NVIDIA AI podcast.
0:00:15 I’m your host, Noah Kravitz.
0:00:17 Since its founding in 1993,
0:00:21 Temenos has been on a mission to revolutionize banking.
0:00:23 Its open platform enables people across the world
0:00:25 to carry out their daily banking needs,
0:00:27 and for banking providers to build new services
0:00:30 and state-of-the-art consumer experiences
0:00:33 using AI and other cutting-edge technology.
0:00:35 Starting a bit more recently,
0:00:37 our guest has been leaning Temenos efforts
0:00:38 to drive digital transformation
0:00:41 from financial institutions across the world.
0:00:44 In October of last year, 2024, to be specific,
0:00:47 Barb Morgan joined Temenos as chief product
0:00:48 and technology officer,
0:00:51 bringing over 25 years of leadership experience
0:00:53 in global product development organizations
0:00:54 with her to the role.
0:00:57 Barb has done a lot in banking and financial services
0:01:00 to put it mildly, especially with AI and cloud tech.
0:01:02 In fact, it’ll be better to ask her
0:01:03 to tell us about her background.
0:01:05 So we’ll start there in just a second,
0:01:07 except that I will add that Barb holds
0:01:09 a Bachelor of Science in Computer Science
0:01:11 from the University of Central Oklahoma.
0:01:13 That said, Barb is here to talk about
0:01:14 generative AI and banking,
0:01:16 Temenos’ approach to AI,
0:01:18 and the importance of sustainability
0:01:19 in the industry for starters.
0:01:20 So let’s get to it.
0:01:23 Bob Morgan, welcome, and thank you so much
0:01:25 for joining the NVIDIA AI podcast.
0:01:27 – Thanks, Noah, excited to be here.
0:01:29 – Excited to have you.
0:01:31 All right, so I teased it in the intro a little bit,
0:01:34 but maybe we can start with you telling us a bit
0:01:35 about your background,
0:01:37 your journey into working with AI,
0:01:39 and how you landed at Temenos.
0:01:41 – Absolutely, so I actually started
0:01:43 with my hands on the keyboard.
0:01:46 So I was a developer many years ago.
0:01:49 When you said 25 years, I had to smile a bit
0:01:51 ’cause it reminds me how long career has been.
0:01:55 But now, my career did start with the hands on the keyboard,
0:01:58 but I always really enjoyed that link
0:02:00 between what we were doing with technology
0:02:02 and how that was really impacting the customer.
0:02:05 And so as my career continued
0:02:07 to kind of go through my journey,
0:02:10 I really gravitated towards those areas
0:02:13 that had a strong customer centricity.
0:02:15 And so I spent about the past 15 years of my career
0:02:18 focused in the financial services industry.
0:02:23 So I’ve led transformations inside banks within techs,
0:02:24 side by side with the banks,
0:02:27 and really focused around modernizing core systems,
0:02:29 building innovative products,
0:02:31 and accelerating AI adoption,
0:02:34 which you can’t have a conversation anymore without AI.
0:02:37 But AI isn’t new to me.
0:02:40 We’ve been using it for years from fraud detection,
0:02:42 risk modeling, automation,
0:02:45 but what’s really different now is we’re seeing that shift
0:02:50 where Gen AI has shifted the entire landscape
0:02:51 where it’s not about efficiency,
0:02:54 it’s really about making the bank smarter,
0:02:58 more intuitive, and bringing that hyper-personalization
0:02:59 to the clients.
0:03:01 It’s exciting when we talk to our clients.
0:03:05 So as you mentioned, I joined Temno’s in October,
0:03:08 and I’ve spent, in my past four months,
0:03:12 a lot of time out there just talking with the clients,
0:03:13 understanding what they’re thinking about,
0:03:15 whether it’s the CEO, CTO,
0:03:19 and they really want to get back to their customers, right?
0:03:22 Whether it’s having us run a banking suite for them on SaaS
0:03:24 so that they can focus on their customers
0:03:28 versus infrastructure, leveraging AI,
0:03:31 but that customer centricity is really coming out,
0:03:33 and it’s paired so nicely with the Gen AI.
0:03:37 – It’s interesting you mentioned the shift from efficiency,
0:03:39 and not that efficiency is a thing of the past,
0:03:42 I’m sure in the banking sector especially,
0:03:44 but kind of that shift from efficiency
0:03:47 to the customer relationship
0:03:49 and how can we better serve customers?
0:03:53 And is that something that you felt happening
0:03:56 before Gen AI in particular
0:03:58 kind of took center stage over the past couple of years,
0:04:00 or is it something that you think
0:04:03 kind of followed the technology that people realized,
0:04:06 like, oh, this is a great way to do all these things
0:04:08 with customer, you know,
0:04:10 personalization, customer service, what have you,
0:04:13 and that trend sort of followed the tech?
0:04:17 – I think the era of the chat box, right?
0:04:20 Which was kind of that first introduction.
0:04:23 When you were sitting on the technology side,
0:04:25 whether inside the bank or at Vintex,
0:04:26 we thought, this is fantastic.
0:04:28 And what we realized was it was actually
0:04:30 really frustrating to the clients.
0:04:33 I don’t know if you’ve ever called into your bank
0:04:37 and it’s like, press one for this, press two for this,
0:04:40 and then you just end up mashing that zero key, right?
0:04:43 – I’m just screaming representative in the phone,
0:04:44 is that right?
0:04:47 And so now I think what we’ve seen
0:04:50 where technology is leading,
0:04:55 is that Gen AI can bring that human centered approach,
0:04:58 and really bringing more humans back
0:05:00 to that immediate touch point with customer,
0:05:05 because now all that data can get pulled instantly.
0:05:07 And so you don’t have to go through
0:05:10 that representative representative.
0:05:14 And so I do think that we thought technology was leading
0:05:16 when kind of that era of the chat bots
0:05:21 and some of the different customer type efficiencies,
0:05:23 we’re playing out maybe five, 10 years ago.
0:05:26 But now I think technology truly is leading
0:05:28 and people are seeing it as an abler
0:05:31 versus again, just that cost efficiency.
0:05:32 – Right, right.
0:05:34 Maybe you can unpack a little bit
0:05:37 what it means from the banker side
0:05:39 to deliver a better experience
0:05:43 and how they’re thinking about leveraging Gen AI
0:05:44 and related technologies.
0:05:46 I love, I’m not even a developer,
0:05:50 but I get on my high horse when people talk about Gen AI
0:05:52 as if it’s the only kind of artificial intelligence, right?
0:05:54 So machine learning, predictive analytics,
0:05:57 all these things, those are not, like we said,
0:05:58 they’re not going away,
0:05:59 but from the bankers perspective,
0:06:01 what are they excited about now?
0:06:04 What are some of the some of us banking customers doing
0:06:07 to leverage this tech to deliver better experiences
0:06:09 and make their clients happier?
0:06:12 – Yeah, I think we’re seeing kind of two themes
0:06:14 really kind of play out
0:06:16 when we have the conversations with the banks.
0:06:20 And the first one is, how can they help their team?
0:06:22 And then secondly, can they help their customers?
0:06:26 So from the perspective of helping their teams,
0:06:27 they’re asking us questions of,
0:06:30 how can we give our bankers instant access to insight?
0:06:33 How can we leverage the historical data
0:06:36 and make recommendations almost instantly?
0:06:38 How can we have those richer
0:06:42 more meaningful conversations enabled for our bankers?
0:06:45 And so that’s really kind of that internal look of his,
0:06:48 how can AI sit side by side
0:06:52 and really be that plus one in the conversation.
0:06:54 I think for the customers,
0:06:56 it goes back to that hyper personalization,
0:06:59 whether it’s pay-learning loan options
0:07:00 or giving them insights
0:07:02 that they hadn’t thought about before,
0:07:06 both from historical, but also predictive in the future.
0:07:09 And then there’s the speed that customers expect now.
0:07:11 Everyone seems to have,
0:07:13 whether it’s chat, GBT or perplexity
0:07:15 or whatever loaded onto their phone,
0:07:17 everyone expects that instant answer now.
0:07:19 And so they expect that everywhere,
0:07:21 but we’re in a highly regulated space.
0:07:23 And so making sure that we do that
0:07:25 in a very responsible way.
0:07:28 But some of the things that we’re doing that,
0:07:31 I really enjoy myself and try to entrench myself
0:07:33 with the teams as much as I can on
0:07:36 is instead of just coming up with solutions
0:07:37 and going out to our clients,
0:07:40 we’re doing a lot of co-design around the AI solution.
0:07:45 So really saying, hey, here’s five or six use cases.
0:07:46 Which of these stands out to you?
0:07:48 Which one or two of these do?
0:07:50 Let’s sit side by side.
0:07:53 Let’s think about how we can co-develop this together.
0:07:54 One that we’re working on recently
0:07:56 is a AI powered solution
0:07:59 that allows the product managers at the bank
0:08:01 to create those financial products,
0:08:04 leveraging that data insight.
0:08:07 And that was where that predictive future capabilities
0:08:08 came into question, right?
0:08:12 So based off of the history, what can we predict?
0:08:14 And we can give you the bank’s knowledge.
0:08:17 We have bank expertise as well,
0:08:19 because we serve 1,000 different banks
0:08:21 or more than 1,000 banks.
0:08:22 And so bringing that all together
0:08:25 to think about future predictions,
0:08:27 pulling it together and giving the right products
0:08:30 to the right customers at the right time.
0:08:33 For me, it’s also,
0:08:35 we don’t wanna leave any of our customers behind.
0:08:38 One of the things that I really enjoy
0:08:40 about what we bring to our clients is,
0:08:41 we focus on flexibility.
0:08:46 And what I mean by that from a pure tech stack perspective
0:08:48 is if you’re gonna be on-prem,
0:08:50 if they wanna be in the cloud or if they want SaaS,
0:08:51 they have the optionality.
0:08:56 And so partnering most recently with Amidia to bring AI
0:09:02 to our on-prem banks has been hugely well received.
0:09:06 Just that ability to give that AI-driven analysis
0:09:09 with those massive data sets that they have on-prem
0:09:12 but allowing them to control the security around it,
0:09:16 allowing them to really not need to have
0:09:19 the deep technical expertise to analyze.
0:09:21 And so a lot of excitement there.
0:09:24 And it’s good too, because we were talking
0:09:26 with the bank the other day and he said,
0:09:29 our ability to have adequate data management
0:09:30 is very limited.
0:09:32 There’s areas we’ve invested,
0:09:35 but it might be 25% of our landscape
0:09:37 that we actually can pull analytics.
0:09:40 So these tools that can look across those massive data sets
0:09:42 are really exciting.
0:09:45 – So to kind of take a step back for a second,
0:09:49 if you don’t mind, Teminos offers a platform.
0:09:51 And so when you’re talking about,
0:09:53 I just kind of wanted to unpack for the listener
0:09:55 what Teminos actually does.
0:09:57 And my understanding is, it’s services,
0:10:01 within you also have a platform where clients
0:10:03 can build products for their banks?
0:10:07 – Yeah, so in kind of three different ways.
0:10:11 So first we have a end-to-end banking platform.
0:10:15 And so for our banks, within like, let’s say within the US,
0:10:17 our tier three regional banks will come in
0:10:19 and they’ll take an end-to-end platform
0:10:22 that provides all of the capabilities of banking.
0:10:25 – Almost like a turnkey banking solution, okay.
0:10:29 And that’s where we see a lot of the adoption of SaaS,
0:10:32 bring us a bank, even within kind of
0:10:33 that neo banking space as well.
0:10:38 But then we also, we also offer modular solutions
0:10:39 to our clients.
0:10:42 And so if you look at some of our tier one banks,
0:10:44 they don’t wanna replace their whole platform, right?
0:10:46 And so, but they may come to us and say,
0:10:50 hey, we just want your payments module
0:10:53 or we just want your originations.
0:10:56 And so giving that choice and flexibility,
0:10:58 whether they want the full platform
0:11:01 or they want modulars within the platform.
0:11:03 And then we also offer products
0:11:05 around what we call point solutions.
0:11:08 So things that may be add-ons that they may choose
0:11:10 to build themselves and plug into our platform,
0:11:11 like their digital interface.
0:11:14 Or we also offer a digital interface
0:11:17 that they can leverage with our suite.
0:11:18 – Gotcha, thank you.
0:11:20 And you serve bank customers of all sizes?
0:11:24 – We do, so we have over a thousand banks globally.
0:11:27 So we have a very global footprint
0:11:30 from the Americas to Mia to APAC.
0:11:34 And so, with that, it’s everything from tier one banks
0:11:37 all the way down to the near banks.
0:11:40 – Are you seeing similar or different trends
0:11:42 in terms of, I guess both adoption rates
0:11:44 from smaller banks and larger banks
0:11:45 when it comes to AI tools,
0:11:47 but also what they’re using them for?
0:11:50 Or is it everyone primarily is customer service?
0:11:53 Like this is the big, it’s not just low hanging fruit.
0:11:55 Like it’s a potentially really big win.
0:11:56 Is that just where the focus is now?
0:12:00 – I would say AI is no longer an option, right?
0:12:03 So they have to have AI, whether it’s embedded,
0:12:07 whether it’s actually more of a feature functionality,
0:12:09 maybe within their digital to allow their customers
0:12:13 to click on some type of AI agent.
0:12:17 But I would say we see two different lens on this.
0:12:21 So one is regionally, you have different areas
0:12:25 that they’re more prone to adopt AI faster
0:12:26 than other regions.
0:12:28 And then from the customer tiering,
0:12:32 what I would say is the tier one, tier two banks,
0:12:34 absolutely they wanted embedded in their product.
0:12:38 The kind of tier three banks, regional banks,
0:12:42 they’re really focused on the personalization that it brings.
0:12:44 – Okay, you kind of alluded to this a little bit,
0:12:49 but is just gathering data and helping the banks
0:12:54 kind of find and gather up and scrub and prepare and use.
0:12:59 You mentioned the example of only 25% of the data
0:13:00 actually going to analytics.
0:13:02 I’m getting that wrong on my paraphrasing,
0:13:04 but is that still kind of the biggest,
0:13:07 I don’t wanna say hurdle, but one of the biggest hurdles
0:13:09 to kind of leveling up success?
0:13:13 – It is, one of the first conversations that I often have
0:13:15 or something I always try to click into
0:13:18 and probably a bit of my kind of geeky background
0:13:21 leads me into is what does your data look like?
0:13:24 Because in order to have responsible AI,
0:13:26 you have to have your data in a state
0:13:29 that you can actually leverage the tools.
0:13:34 And for us, we really take that AI explainability
0:13:35 very seriously.
0:13:37 And so we spend time with our customers
0:13:40 to understand what is the state of their data.
0:13:43 And oftentimes what we find is if you think back
0:13:45 to the transformation 10 years ago
0:13:47 that banks were undergoing,
0:13:51 it was a lot about how it looked and felt to the customers,
0:13:53 not about transforming the back end.
0:13:57 So many banks are made up of acquisitions,
0:13:58 mergers over time.
0:14:02 And so the front end looks really slick and great
0:14:04 and it looks like you’re dealing with one bank
0:14:06 and it’s actually hitting six different banks
0:14:07 in the background.
0:14:08 Right.
0:14:13 And so sometimes it is, working with our clients,
0:14:17 we have a Temnos data product, our RTDH product
0:14:22 and it is bringing that data into a state
0:14:24 that it can then be leveraged.
0:14:27 And so kind of back to that optionality
0:14:29 that we talked to our clients about,
0:14:32 sometimes we may just take a portion of the bank
0:14:36 and really focus on getting the data streamlined
0:14:39 and getting it ready to be able to use that embedded AI
0:14:40 and then proving it out.
0:14:44 And I think as we see, banks are highly regulated,
0:14:45 that’s not going away,
0:14:49 the regulations are gonna get tougher if not anything else.
0:14:52 And so we always keep that top of mind,
0:14:54 making sure that compliance standards
0:14:58 are absolutely embedded into our products
0:15:00 that fully auditable.
0:15:02 And so that’s where a lot of our conversations
0:15:04 end up leading around data is, are you ready?
0:15:05 Right, right.
0:15:07 I’m speaking with Barb Morgan.
0:15:11 Barb is the chief product and technology officer
0:15:14 at Temnos, a role she started within the last year,
0:15:17 the current chapter in an illustrious career
0:15:19 in global product development,
0:15:21 particularly across banking and financial services.
0:15:24 And we’ve been talking about generative AI in particular
0:15:27 and the banking perspective on the chat bot revolution,
0:15:29 if you wanna put it that way,
0:15:30 but just as a leaping off point,
0:15:34 so much more obviously before chat GPT and the bots came
0:15:37 in the world of machine learning and obviously sense
0:15:40 the pace, it’s just been breakneck.
0:15:42 Barb, I wanna kind of shift gears for a second here
0:15:45 if we can and talk a little bit about sustainability
0:15:46 in the industry.
0:15:49 I know that Temnos has a point of view on this
0:15:52 and is actively working with client organizations
0:15:53 to help them be more green.
0:15:55 And I would think that your perspective,
0:15:58 both in the industry and also in kind of the international
0:16:01 nature of the work that you do with Temnos,
0:16:04 curious to hear both the company’s perspective
0:16:07 and what you’re up to with clients to help them be more green,
0:16:10 but also your take, having been around the industry
0:16:13 and around the world literally for a few decades now.
0:16:17 – So to me, sustainability, it’s not a trend.
0:16:19 I think when it really started becoming part
0:16:22 of the conversations, call it maybe 10 years ago
0:16:25 where it actually was part of annual reports
0:16:28 and things like that, I think people kind of questions,
0:16:30 like is this long-term?
0:16:31 And now what we’ve seen the shift is,
0:16:34 it’s really a responsibility of the organization.
0:16:38 And so, when we sit down and we talk with our banks
0:16:39 and why it’s important to us
0:16:42 and bringing some of those solutions forward
0:16:45 that allow them to be greener,
0:16:48 we talk about both what it means to them,
0:16:49 where their focus is.
0:16:52 So to your point, we serve banks globally
0:16:54 and so we see different parts of the world
0:16:59 where it may be more around their carbon accounting
0:17:01 or some areas that may be like,
0:17:04 hey, we really want to understand
0:17:06 how our cloud deployments are helping.
0:17:08 And so understanding like,
0:17:11 how can we drive a greener banking future?
0:17:14 And there are always great conversations
0:17:17 because they really often talk to the values
0:17:18 of the organizations.
0:17:20 And so you get to actually spend time
0:17:22 in the cultural side of the bank.
0:17:26 And what’s really kind of cool for the lack of a better word
0:17:30 about those conversations is when you can use AI
0:17:32 to bring the culture forward in the solutions
0:17:34 through something that really matters to them,
0:17:37 it’s a very rewarding solution, right?
0:17:41 So, like I talked about the smart carbon accounting,
0:17:44 helping our customers track their carbon footprint
0:17:47 both through how they’re using their software
0:17:49 but then being able to offer that to their customers.
0:17:51 There’s many consumers who want to know,
0:17:54 like, hey, how are the purchases that I’m making
0:17:56 impacting my carbon footprint, right?
0:17:58 So not only are we talking to our customers
0:18:00 but we’re actually impacting their end customers as well.
0:18:04 – So kind of to piggyback off that a little bit
0:18:06 and open it up a little more abstractly, I guess.
0:18:07 This is a big question,
0:18:09 but I’ll throw it at you, you can handle it.
0:18:13 How do you see the future of banking being shaped by AI?
0:18:15 And I guess the flip side of that is,
0:18:17 how do you see the future of AI growing and banking?
0:18:21 But I think really, there’s been so much,
0:18:22 we kind of joked about it for a second.
0:18:24 There’s been so much in the past couple of years
0:18:26 with gen AI and the pace isn’t slowing down.
0:18:30 And we have in our notes here to talk about AI agents,
0:18:32 which is kind of the latest thing
0:18:34 buzzword-wise in the past few months, right?
0:18:35 But certainly not a new thing
0:18:37 and certainly something that could wind up
0:18:40 really shaping the technology going forward.
0:18:41 We’ll have to see what plays out.
0:18:43 Do you have a strong view on what you think
0:18:47 is going to happen with AI and banking
0:18:49 and banking kind of being reshaped by the technology?
0:18:52 And that can be short-term next couple of years,
0:18:54 take it a little further out if you like.
0:18:56 What are your thoughts kind of generally on this?
0:18:59 – Yeah, I think even before stepping into that,
0:19:02 I think the one thing that I really see
0:19:06 and I think it’s important to kind of talk about
0:19:09 is the leadership of an organization
0:19:13 really shapes how AI is going to be accepted, right?
0:19:16 Is it the same as a friend or a foe?
0:19:18 If you have your top leadership just talking about
0:19:21 how much money we’re gonna save from AI, it’s a foe, right?
0:19:25 But we’re seeing the leaders of the organizations
0:19:28 really look at AI as not as a threat
0:19:32 and really talking about it as an enabler,
0:19:35 getting people curious, getting people engaged,
0:19:38 more and more organizations, and we do this ourselves,
0:19:40 whether you use the term eating their own dog food
0:19:44 or French would say drinking their own champagne.
0:19:46 We’ve been doing that ourselves to say,
0:19:49 “Hey, let’s actually use this on ourselves.”
0:19:51 And then if it works well for us, great.
0:19:53 We can start to expand it to our customers.
0:19:56 And so when you start to see the leadership
0:19:58 of the organizations, whether it’s the CEO
0:20:02 or any of the C-suite, talking about how they’re curious,
0:20:04 how they’re using it in their daily lives,
0:20:08 how they’re getting in there and playing around themselves
0:20:11 and thinking about how can I get rid of repetitive,
0:20:14 time-consuming tasks and focus on deeper matters
0:20:15 and more strategic work,
0:20:18 you start to see that really come out.
0:20:20 And I think that’s important in order
0:20:22 for people to see AI as a tool
0:20:26 to amplify the human potential, not to replace it.
0:20:30 – Are the bankers, the employees, to put it that way,
0:20:32 are they thinking about it the same way?
0:20:33 Is there excitement?
0:20:35 Is there fear around job replacement?
0:20:37 Is I think– – Any change, right?
0:20:39 There’s always gonna be a bit of fear.
0:20:44 And I think it’s up to us as the banking experts
0:20:46 and as partners to our clients
0:20:49 and then working with their teams
0:20:52 to help kind of show how the change does
0:20:54 actually help them, right?
0:20:59 And so when we see that kind of pivot away from,
0:21:01 “Oh my gosh, this is gonna replace me,”
0:21:02 to, “Wait a second,
0:21:05 “I’m actually gonna sit side-by-side with AI.”
0:21:08 And it’s gonna– – Plus one, you–
0:21:09 – Yes. – I forgive me in a rub.
0:21:10 I just wanted to give you credit.
0:21:12 I hadn’t heard somebody use Plus One
0:21:14 to talk about AI before and I love it, right?
0:21:17 It’s the co-pilot, whatever you wanna call it,
0:21:18 but Plus One is great.
0:21:22 – Yeah, and for me, I really focus on,
0:21:25 I hate the word artificial intelligence
0:21:28 because artificial, it’s fake.
0:21:30 There’s just that negative connotation.
0:21:33 And so I often start out by talking with our clients
0:21:36 about thinking about it as augmented intelligence.
0:21:40 And that gives you that Plus One effect, right?
0:21:43 And then when you show the bankers,
0:21:46 hey, someone’s gonna walk into your branch,
0:21:48 you’re instantly gonna be able to know
0:21:51 more about that customer than they know about themselves.
0:21:55 And you’re gonna be able to have a really deep conversation
0:21:57 both about what’s right for them today,
0:22:00 what’s right for them in the future,
0:22:02 how they’re shaping those things.
0:22:03 Their eyes light up, right?
0:22:06 Because oftentimes they would have to,
0:22:08 the customer would sit in the lobby,
0:22:09 they would do a bunch of research,
0:22:11 they might be pulling paper files out,
0:22:12 they’re trying to remember,
0:22:15 “Okay, this person has been with us for 10 years
0:22:17 “and they have a mortgage and they have a car
0:22:19 “and they have this.”
0:22:19 – Right, right.
0:22:21 – “Oh gosh, what else could I offer them?”
0:22:24 When they can, through natural language questions, say,
0:22:27 “How long has this customer been with us?
0:22:29 “What is their familial history?”
0:22:33 So this may be a 30 year olds,
0:22:37 but maybe their family’s been with the bank for 25, 30 years.
0:22:40 And then when their customer walks up and they say,
0:22:44 “Hey, it’s been great, we’ve served your parents.”
0:22:47 And so excited to have you here with us.
0:22:49 And we had these great,
0:22:51 we’re looking at what loans that you have with us,
0:22:53 we could consolidate those together,
0:22:54 we could offer you a better rate.
0:22:57 We have this great potential over here.
0:22:58 They’re excited, right?
0:23:01 – Yeah, it really kind of,
0:23:03 you sort of made real in listening to you talk about that.
0:23:04 It’s a great example,
0:23:07 ’cause it makes me think of the sort of abstract talk
0:23:11 about machine learning, AI tech, kind of,
0:23:14 freeing humans up to do what humans do
0:23:15 better best.
0:23:18 And in this case, I can relate because it’s not quite AI,
0:23:21 but if it’s not in my phone calendar, I forget it, right?
0:23:24 And so I can only imagine being a banker,
0:23:26 having so many clients to serve.
0:23:28 As you said, I’m in the lobby waiting
0:23:30 because the banker is doing their best
0:23:33 to kind of do a crash course on my whole history
0:23:36 with the bank to serve me, ’cause there’s so many customers,
0:23:37 I’m getting frustrated and waiting, et cetera.
0:23:39 Yeah, let the AI do it.
0:23:41 And then it just, in real time, it pops up.
0:23:43 And yeah, that’s a great example.
0:23:47 And today, I mean, the future is gonna be all
0:23:50 about human and AI collaboration.
0:23:52 We’re already seeing kind of AI agents.
0:23:54 So to your agentic AI, right?
0:23:58 And latest buzzword, handling those routine banking tasks.
0:24:02 But if you can set it up where it’s doing your segmentation,
0:24:04 then it’s doing some product suggestions,
0:24:07 then it’s seeing, as you offer those products,
0:24:10 maybe it’s actually shaping that segmentation.
0:24:12 And so those agents are continually learning
0:24:13 from each other.
0:24:17 And then you can bring that to that human collaboration.
0:24:19 It’s just exciting.
0:24:23 I think we’ll start to see digital humans in making
0:24:26 so that you aren’t saying representative representation.
0:24:29 I was gonna say, when we were talking about that before,
0:24:33 I’m 100% for the plus one,
0:24:35 the augmented intelligence is also,
0:24:37 I like that way of thinking about it as well, right?
0:24:41 And I wanna see humans use the technology to thrive
0:24:43 and not talk about things like replacement, et cetera.
0:24:46 That being said, I’m an impatient person.
0:24:49 And so I always gravitate towards the self-checkouts
0:24:51 at stores.
0:24:55 And so if the automated banking menu could give me,
0:24:58 ’cause I never call unless I have some weird question
0:25:00 or like I’ve missed four payments
0:25:03 and wanna try to beg somebody to give me grace, right?
0:25:05 So yeah, get the automated system to that point
0:25:07 and I’ll be happy.
0:25:10 Yeah, or even think about how great would it be
0:25:13 if your phone pops up, ’cause I know we all,
0:25:16 or at least I know my phone is always within an order.
0:25:17 Yes, yep, yep.
0:25:19 Anywhere, and if it just said,
0:25:22 “Hey, Noah, looks like you missed your last payment.
0:25:24 “Would you like us to auto debit from your account
0:25:27 “and we’ll free up any late fees?”
0:25:28 And you’re like, “Yes, done.”
0:25:30 Yeah, that’s exactly the one, yeah.
0:25:34 And so that proactive monitoring, bringing that,
0:25:36 so that you aren’t even having to call in, right?
0:25:38 You, how much better would that be
0:25:41 if it automatically reaches out to you?
0:25:45 Yeah, if my creditors happen to be listening to the podcast,
0:25:48 I just made that example up, we’re good.
0:25:51 Barb, before we get to wrapping up, I wanted to ask you,
0:25:53 and I think this is something I need to start asking guests
0:25:56 going forward, so thank you for inspiring me.
0:26:00 You mentioned in your work talking to leadership,
0:26:03 it’s so important, it’s such a tone for so many things,
0:26:06 but including an organization’s a bank’s kind of perspective
0:26:07 on embracing AI.
0:26:10 And you talked about getting these clients,
0:26:12 these banking leaders, to start using the tools
0:26:12 in their own life.
0:26:13 Do you have a routine?
0:26:16 Do you have things that you use AI for
0:26:19 on a daily, regular basis that,
0:26:22 maybe a pro tip to share with the audience?
0:26:27 Yes, I might overuse it, that’s probably the engineer in me.
0:26:30 My husband, I’ll be traveling and I’ll get a message,
0:26:31 like, can you turn this thing off?
0:26:33 Like, why are all the lights coming on?
0:26:36 Why is, like, I know you wake up at six,
0:26:38 but I’m not waking up at six when you’re not home.
0:26:40 Like, why is the house waking up for me?
0:26:41 Smart home lights, yeah.
0:26:43 No, no, I mean, you know, for me,
0:26:45 I use it in a couple different ways.
0:26:49 Sometimes I use it just to say, is my message clear, right?
0:26:54 Like, when you’re so deep into whatever your specialty is,
0:26:55 right?
0:26:57 You feel like your message is clear
0:26:59 because you’ve been living, eating, breathing,
0:27:00 working on it for a while.
0:27:03 I can quickly throw that into whatever my favorite tool is,
0:27:06 whether it’s chat, GBT, or perplexity, or whatever,
0:27:10 co-pilot, and say, summarize this message.
0:27:12 What is the tone?
0:27:16 What level of audience is this reaching?
0:27:20 And hopefully, it’ll say, hey, this is actually
0:27:22 geared at a engineering audience.
0:27:24 Oh, well, wait a second, that’s not who I’m speaking to.
0:27:25 Right, that’s great.
0:27:29 Let me make sure that I bring this back into more
0:27:33 of a business speak, or this is very financially focused.
0:27:34 OK, wait a second.
0:27:37 And so I use it oftentimes in a way to sense, check me.
0:27:40 But I also use it for a bit silly stuff, right?
0:27:45 So we have four kids back in the state, all college age,
0:27:47 nine to 22.
0:27:50 And we were going on holidays in Mexico.
0:27:53 And as much as I think I’m a cool mom,
0:27:56 I absolutely used AI to say, what are the best things
0:27:57 to do down in Mexico?
0:27:58 Totally.
0:28:01 Yeah, and it got it pretty close to right.
0:28:03 Like, they like the different restaurants
0:28:04 that we took them to.
0:28:05 And there you go.
0:28:07 Yeah, so I use it quite often.
0:28:11 I also– I play with a lot of tools
0:28:13 outside of the financial industry,
0:28:16 because I think it’s important to see how other industries are
0:28:16 leveraging AI.
0:28:19 It gives us ideas into the financial space,
0:28:21 whether it’s maybe the insurance space.
0:28:23 I was on my insurance app the other day,
0:28:25 and they have AI embedded.
0:28:27 And I thought, wow, this is really cool.
0:28:32 And so looking for ways that other people are using AI
0:28:34 is sometimes the way that I use AI.
0:28:35 Excellent.
0:28:38 Barb, for listeners who would like to learn more
0:28:43 about Temenos approach to AI, other services Temenos offers,
0:28:47 maybe something a little more engineering, geeky oriented.
0:28:48 I don’t know if there’s a developer blog,
0:28:51 or you have other social media, anything.
0:28:54 Where would you direct them to go after listening?
0:28:56 Yeah, whether it’s LinkedIn, if that’s
0:29:00 their favorite within just looking at Temenos.
0:29:03 They will definitely find a cluster of areas to go.
0:29:09 And then, of course, our website, just our www.temenos.com.
0:29:10 They can look at our products.
0:29:13 We do have more of the technical aspects, right?
0:29:16 So our developer portals, and then also just understanding
0:29:19 where our thought leadership is in the space.
0:29:19 Fantastic.
0:29:22 Barb, Morgan, thank you so much for joining the podcast.
0:29:24 This was a pleasure.
0:29:26 I learned some things, which I knew I would.
0:29:27 We talked before we started.
0:29:30 Banking’s not my wheelhouse, so I appreciate that.
0:29:30 Thank you.
0:29:32 But more so, it’s just kind of– it’s always
0:29:36 fascinating to talk to somebody who’s a leader in their field
0:29:39 and has been living and breathing it for long enough to–
0:29:42 we’re talking about world-changing technology,
0:29:44 but there are deeper things that have been around for a while
0:29:48 now that are really important to shaping your perspective.
0:29:50 So your perspective is greatly appreciated.
0:29:52 Thank you.
0:29:55 [THEME MUSIC]
0:29:59 [THEME MUSIC]
0:30:02 [THEME MUSIC]
0:30:06 [THEME MUSIC]
0:30:09 [THEME MUSIC]
0:30:13 [THEME MUSIC]
0:30:17 [THEME MUSIC]
0:30:20 [THEME MUSIC]
0:30:24 [THEME MUSIC]
0:30:28 [THEME MUSIC]
0:30:31 [THEME MUSIC]
0:30:35 [THEME MUSIC]
0:30:38 [THEME MUSIC]
0:30:42 [MUSIC FADES OUT]
0:30:52 [BLANK_AUDIO]

0:00:10 [MUSIC]
0:00:13 Hello, and welcome to the NVIDIA AI podcast.
0:00:15 I’m your host, Noah Kravitz.
0:00:21 On October 17th of 2024, NVIDIA was honored to host the Technovation World Summit
0:00:24 finalist pitch and awards ceremony in Santa Clara, California.
0:00:28 The summit featured more than 50 young innovators from around the world,
0:00:30 pitching tech solutions for sustainable development goals.
0:00:33 And it was a celebration of what happens when the creativity and
0:00:37 resilience of girls meets the transformative power of technology.
0:00:41 Here to talk about the World Summit, Technovation, and
0:00:44 the incredible work young women are doing across the globe to tackle our
0:00:47 biggest problems are Tara Choklovsky and Anshita Sani.
0:00:53 Tara is the founder of Technovation and a repeat guest on the podcast,
0:00:56 which I’m terribly excited about, we’ll get into that in a second.
0:01:01 And Anshita is an alumna of Technovation and currently works as a member of
0:01:06 the technical staff at Open AI, an AI company some of you might have heard of.
0:01:09 Tara, Anshita, thank you so much for taking the time to join the podcast.
0:01:12 Welcome, congratulations on the awards summit.
0:01:14 That was a few months ago.
0:01:16 So I hope your new years are off to a great start and welcome.
0:01:17 >> Thank you, Noah.
0:01:21 NVIDIA has been such a long-term supporter of Technovation that I’m so
0:01:25 happy to be back and so happy to share our stories along with.
0:01:27 An alumna we’re super proud of, Anshita.
0:01:31 >> Well, I am, like I said, Tara, I’m thrilled to get to talk to you again.
0:01:35 And I’m equally thrilled to meet you, Anshita, and to hear about the work you’ve
0:01:35 been doing.
0:01:37 But let’s start, Tara.
0:01:40 You were on the show about five years ago, Time Flies.
0:01:44 At that time, you were the founder of a nonprofit called Iridescent.
0:01:47 You were described, we use this quote from somebody else, not our words.
0:01:50 But you were being described as a pioneer empowering the incredible tech girls of
0:01:51 the future.
0:01:52 Seems like prescient words.
0:01:57 What have you been up to since then and how did it all lead to now to starting
0:01:59 Technovation and us getting to talk to you again?
0:02:00 >> Yeah, thank you.
0:02:03 I think I sort of see Iridescent initially.
0:02:07 That was the name of the organization, the nonprofit that I started almost 19 years
0:02:12 ago with the goal to empower underrepresented communities with the
0:02:18 skills that they could be the innovators of technologies that solve the problems
0:02:22 they face so they didn’t have to wait for a savior.
0:02:28 And so I think when NVIDIA first supported us back in 2018, I could sense
0:02:33 that AI was going to be the most powerful revolutionary force.
0:02:37 And it was interesting, like we started to do that actually in 2016.
0:02:42 And I was talking to people and I was like, we need to revamp our curriculum to
0:02:43 focus on AI.
0:02:47 And nobody was talking about that, especially not in the education space.
0:02:51 And everybody would be like AI and underserved communities don’t go in
0:02:53 the same sentence.
0:02:57 And NVIDIA was one of the first companies to actually pay attention to that.
0:03:01 And they said that, okay, we will fund this program.
0:03:06 And so we launched the first AI family challenge back in 2018, 2019.
0:03:09 And it was the first global AI education program in the world.
0:03:16 And everybody, there was so much negative feedback that I got that this will not work.
0:03:17 People don’t care about it.
0:03:21 And we actually opened it up and we went into communities.
0:03:26 I saw we had a refugee community in Somalia participate in the AI family challenge.
0:03:30 We had low income communities from Michigan participate families.
0:03:33 And they all came to me and they were saying that, thank you for teaching us
0:03:37 about how these things work because we see it in our phones, we see it everywhere.
0:03:41 But nobody ever bothered to tell us about how it works.
0:03:44 And with NVIDIA, we did these three design challenges.
0:03:49 That was literally in 2019 on how a self-driving car works, how a neural network works,
0:03:53 and how parallel processing works, like using hands-on experiments.
0:03:56 Those videos are still some of our top videos of all time.
0:04:03 Because guess what, neural networks, people actually know what LLMs stand for and all that kind of stuff.
0:04:08 So I’m just so excited that NVIDIA had such a huge role to play in pioneering.
0:04:11 Not just AI, but also AI education.
0:04:15 And then when we were looking at the data of what works, what doesn’t work,
0:04:19 the program that was really coming to the forefront that has long-term impact on
0:04:24 participants’ identity was this technology model of girls going through an accelerator.
0:04:29 A three-month accelerator where you find problems that you care about and actually build AI solutions
0:04:34 or mobile app solutions and pitch to investors and have a demo day.
0:04:39 And the whole experience of working in a team, being the founder, is just transformational.
0:04:44 And so we changed the name of the organization from iridescent to technology.
0:04:47 We cut all our other programs and just focused on this accelerator.
0:04:52 And then everybody else started to recognize AI was a thing.
0:04:57 And then in 2022 when ChatGPT came out in November 2022.
0:05:02 So we were the first education organization to remote and we developed a whole curriculum
0:05:06 for our girls to say, this is how you use ChatGPT to help with identification,
0:05:08 with help with developing your code.
0:05:15 And then we had thousands of girls in December 2022 using ChatGPT to sort of come up with,
0:05:16 help them in their accelerator journey.
0:05:18 And since then, of course, I took.
0:05:24 I’m just remembering so vividly that this was around the time or was the time when a lot of
0:05:29 school districts and educators, and I mean, no shade in saying this, were really struggling.
0:05:30 Well, what do we do with this?
0:05:33 We need to ban it so we can figure out how people won’t cheat.
0:05:37 And here you are saying, no, no, no, no, let’s people want to know how this works.
0:05:37 Let’s embrace it.
0:05:38 Let’s use it.
0:05:39 And yeah, absolutely.
0:05:44 So, you know, I remember going to a local school district here in Silicon Valley and telling
0:05:48 them, OK, you’ve got to use ChatGPT and they were like, what is this?
0:05:52 And so they’re like, oh, it blows their mind.
0:05:54 And they’re like, we need to get permission from.
0:05:56 I’m like, it’ll be too late, right?
0:06:00 Get this, have the girls use it and then you can deal with the consequences later on.
0:06:03 But anyway, so and then I mean, there’s more and more information.
0:06:06 About how AI is going to revolutionize the world, right?
0:06:11 And so I was like, the problem of not having enough people developing these
0:06:15 technologies is only going to get bigger because these technologies are moving so fast.
0:06:19 And so we launched the AI Forward Alliance with UNICEF as a core partner.
0:06:24 And the goal was, and the goal still is to empower 25 million innovators, AI
0:06:28 innovators, and especially females, because guess what?
0:06:32 Like globally, there are only, I think, 18 million tech professionals in the world.
0:06:35 Only 14 million men and only 4 million women.
0:06:37 I do this all the time in the show.
0:06:38 And but it’s genuine.
0:06:39 I’m just not a big stats person.
0:06:43 So I apologize, but unfortunately, the 14 four split doesn’t surprise me.
0:06:45 It’s yeah, we need to work on that.
0:06:47 Yeah, goes without saying.
0:06:48 But is that only 18 million?
0:06:50 Wow. OK. Yeah, it’s less than half a percent.
0:06:55 So the reason is so I did this study with the ILO workforce data and we went country
0:06:57 by country by country, whichever country has data.
0:07:01 And I think the reality is like, I mean, if you look around, maybe not in Silicon Valley,
0:07:05 but if you look around, most people are employed in the service sector in the world.
0:07:09 Right. And so people who can actually build technology, they’re very few.
0:07:13 And so I think if countries need to progress economically, they actually
0:07:14 need to have a bigger talent base.
0:07:16 And where are you going to get your talent base?
0:07:17 Happen to the one that’s not part of it.
0:07:21 Right. So so that’s that’s that’s where the AI forward alliances.
0:07:24 And last year, we were just so we had like record numbers.
0:07:28 Like I think we had like 35,000 girls participate in the Technovation Program.
0:07:32 And in the finalists, it was awesome to be able to bring them to Nvidia.
0:07:36 I was and I’m now feeling it again, so bummed that I couldn’t make it down in person.
0:07:38 But by all accounts, it was a terrific event.
0:07:42 Do you want to tell us about some of tell us about the summit?
0:07:44 Do you want to get into some of the pitches?
0:07:46 Yeah, I can. It’s just interesting, right?
0:07:50 Like when you give technology to people, especially young people,
0:07:53 that’s when you see like the limits of the points of breakage, right?
0:07:56 Track technology, but also where innovation comes.
0:08:01 And I think what I’ve always seen is like young people, especially girls,
0:08:05 because you underestimate them all the time, they just have such interesting ideas.
0:08:06 So I’ll give an example.
0:08:11 In 2012, I had a team of girls come up and create a mobile app
0:08:14 that was around focusing and mental health.
0:08:18 This was 13 years before this whole mental health and wellness.
0:08:20 And I could not understand that.
0:08:24 And I was like, why on earth would you want an app that is blocking other apps?
0:08:28 Because this increases productivity and I actually give them negative feedback.
0:08:32 OK, because I was it was so against what I was seeing.
0:08:36 Yeah, but they were like, no, this is really messing with my productivity
0:08:39 because I’m not so addictive and it’s messing with how I feel.
0:08:41 And I was like, wow, right?
0:08:44 Like this was literally 13 years before the adults caught up.
0:08:46 And so it’s a similar thing like this year.
0:08:51 We had such interesting solutions where we had some Vietnam,
0:08:56 which are about preserving their cultural heritage through lullabies
0:08:58 and putting them into large language models.
0:09:01 Also sort of Vietnamese facial expressions,
0:09:04 because there are a lot of the facial expression recognition systems
0:09:06 don’t recognize Vietnamese facial expressions.
0:09:09 So they did their own training of data sets.
0:09:12 And so I see sort of the technology community as like
0:09:17 bringing last mile data to large language models or large models
0:09:20 and helping them become more representative of the world.
0:09:23 As we record this, the episode that just went live
0:09:27 was an episode with the founders of a company called GUI AI,
0:09:31 and they are doing conceptually, to me, very similar things and,
0:09:36 you know, feeding localized data around populations, communities,
0:09:39 specific problems, challenges to be solved into LLMs
0:09:43 to create these tools to actually serve the users in those places.
0:09:47 Just the whole idea of, you know, you’re wherever you’re from,
0:09:50 wherever you live, your community’s knowledge, your language,
0:09:53 your history, your stories, your best practices of doing the things
0:09:57 that you do to thrive and survive, you know, that’s just the most important data.
0:10:02 And so being able to use that positively to, you know,
0:10:04 so I think it’s an area that I’m hoping gets more and more exposure
0:10:07 because it’s really, I mean, it’s vital, it’s important.
0:10:09 And it raises the quality for everyone, right?
0:10:11 Because the large language model becomes better.
0:10:14 Just it’s not boring anymore, it doesn’t show in this corporate.
0:10:15 Yes. Yes.
0:10:18 I think I remember one long researcher was saying his wife
0:10:20 calls chatty-pity-chatty boy.
0:10:23 Oh, I’m sorry I took that one.
0:10:25 But I think those are the things that we need to change, right?
0:10:28 And I think that’s what this community is thinking to change, so.
0:10:32 I’m speaking today with Tara Chaklowski and Sheeta Saini.
0:10:36 We were just speaking with Tara about her organization Technovation
0:10:39 and the origin story from when we had Tara.
0:10:42 I didn’t say this earlier, I should have, but it’s episode 85.
0:10:46 If you want to go back in the archives and listen to Tara’s first appearance
0:10:49 on the show, Teaching Families to Embrace AI.
0:10:53 And we’ve been talking now about Technovation and the origin story
0:10:58 and the mission of it to, you know, really unlock the innovators of that we need today.
0:11:03 We certainly need going forward in the future and girls, females, women in particular.
0:11:08 And so now and Sheeta would love to hear your story of how you got involved with Technovation.
0:11:12 You don’t have to be the sole spokesperson for the Technovation experience,
0:11:16 but I really delighted to have you here with us today and would love to hear your story.
0:11:20 Yeah, for sure. So I started getting involved in Technovation through,
0:11:23 I think just serendipity, really lucky to have found the program.
0:11:27 But I had just taken my first computer science class in high school.
0:11:30 That was just AP computer science.
0:11:34 And we had done a lot of fun projects in that class that sort of showed me what coding meant,
0:11:40 what it means to create a program using, you know, just your laptop and your fingers.
0:11:44 And so from there, I was thinking, all right, this is really cool,
0:11:46 but what can I actually do with this?
0:11:48 Right? We made programs like Boggle and stuff like that.
0:11:53 But I wanted to see if I could do something actually useful and in the real world with it.
0:11:56 And so I was looking around for different opportunities to that end.
0:12:01 And that’s how I came across Technovation and that whole mission of empowering girls
0:12:05 to solve problems in their community was something that really resonated with me.
0:12:09 And so I actually ended up starting a club in my high school Technovation Club.
0:12:13 And we were just bringing together a bunch of girls from across our high school
0:12:17 that were interested in this program and interested in finding a problem to solve.
0:12:21 And so one of the first problems that I tackled through Technovation
0:12:25 actually ended up being really personal to me and really personal to the situation
0:12:28 that was going on in my high school at the time that was vaping addiction.
0:12:33 And so this was happening at my high school, I think maybe 2018, 2019.
0:12:37 And in fact, I think the situation became so bad
0:12:40 that we actually had the doors taken off of the bathrooms at my high school.
0:12:44 And so when I sat down at Technovation to think about problems
0:12:47 going on in my community with a group of friends, we said, look,
0:12:48 this is going on right around us.
0:12:53 What can we do with computer science with technology to help this situation?
0:12:59 And so we ended up going through a lot of research from different Johns Hopkins studies.
0:13:02 We had mentors actually from Microsoft that helped us
0:13:04 through the process of building this app.
0:13:09 And I think at the end of it, it’s just really empowering to have that concept of,
0:13:13 hey, like Tara said, I can look at these problems and I can solve them.
0:13:17 I don’t have to wait for somebody else to jump in and solve this problem.
0:13:19 And so that’s what’s got involved with Technovation.
0:13:22 Amazing. I have to know, though, what was the app?
0:13:24 What was the solution that you came up with?
0:13:26 Yeah, so we, again, looking at those different research studies,
0:13:29 we came up with sort of a couple of different approaches.
0:13:31 So I shouldn’t say solution.
0:13:33 What’s the approach? That’s the question. Yeah.
0:13:36 Yeah, there was a couple of different features that we had in our app.
0:13:37 One thing was pretty simple.
0:13:41 It was a craving control timer, which actually was to be remarkably effective.
0:13:45 So setting a 10 minute timer when you have the urge to pick up a vape
0:13:49 or pick up some addictive thing is very helpful to curb that craving.
0:13:53 We would also take a user’s profile of interesting hobbies,
0:13:57 other things that they like to do and suggest those at the time of the craving
0:14:01 control timer. Right. So you’re just like to rail the craving into something positive.
0:14:03 Yeah, so simple, but it’s so hard for humans.
0:14:07 Yeah, yeah, I think I still need that to write with my phone.
0:14:12 So yeah. And then another aspect that we had was trying to create
0:14:15 a personalized quitting timeline that you could visualize along a graph.
0:14:19 We had another chat system so that you could chat with your support system
0:14:20 that was, you know, helping you through this process.
0:14:23 I think addiction is not just something that you can solve in an app.
0:14:27 It’s something where you need people, you know, a support system in community.
0:14:29 So. So a chat system hooked up to humans.
0:14:31 Yes, exactly. Yeah, amazing.
0:14:34 Yeah, with your friends, they would be on the app as well to, you know, got it.
0:14:36 OK, right, right, right.
0:14:38 That sounds amazing and not to diminish it all.
0:14:40 But you’re still in high school at that point.
0:14:42 Now I make sure we have time to keep going forward.
0:14:46 So that was your first Technovation experience, building that up incredible.
0:14:50 And so you’re in high school, you’re 14, 15, something like that.
0:14:52 Maybe how does the story progress?
0:14:56 Right. So I think, again, after Technovation, it was I think that was sort
0:14:59 of the aha moment in computer science for me, where it wasn’t some gimmick.
0:15:02 It’s it’s something that people are using every day to solve these problems.
0:15:05 And so I wanted to go further down that path.
0:15:08 And especially I was interested that in that time in these health things,
0:15:11 after after building this vaping app, I was wondering what what are other problems
0:15:14 that I can go after in this, you know, mobile health care space.
0:15:17 Yes, well, I ended up going out for an opportunity that
0:15:21 well, I never thought I would land it, but it was this research internship
0:15:24 at the University of Washington in Seattle, in Seattle. Yeah.
0:15:27 I’ll go Huskies, my wife’s an alum.
0:15:29 Oh, really? OK. Yeah, I love you, Doug.
0:15:32 And so, yeah, so I applied to this opportunity at the ubiquitous computing
0:15:36 lab, it was called at the University of Washington, and they had a lot of
0:15:39 different projects around solving what seemed to me at the time really big,
0:15:43 complex problems with the help of computer science or just a mobile phone
0:15:44 or something like that.
0:15:47 And so the project that I ended up working on that summer at that lab
0:15:54 after Technovation was essentially being able to detect, count and classify
0:15:57 different exercises with just a phone’s microphone.
0:15:59 And so that sounds kind of hard to wrap up.
0:16:04 So people doing physical exercise, but you’re using the mic, not the camera,
0:16:08 just the mic to detect what they’re doing and then do something with that information.
0:16:10 Exactly, right. That sounds super cool.
0:16:14 Yeah, it was I think it was really cool from a technological perspective,
0:16:16 but also for who we were solving the problem for.
0:16:19 Before you get into that, can I just ask, because again, I got it,
0:16:21 it’s the NVIDIA podcast I got asked.
0:16:25 So were you like bouncing sound waves off from the microphone or?
0:16:30 Yeah, exactly. The whole detection and everything just came from like the Doppler effect.
0:16:31 Yeah, yeah, yeah. Yeah, that’s so cool.
0:16:35 When you’re doing exercises, the waves just reflect back to the phone in a different way.
0:16:40 And so if we just play a really high frequency sound, we can’t even hear it, right?
0:16:43 But the phone is able to pick up the reflection of the sound back to the phone.
0:16:48 And so from there, you can actually see really interesting patterns in the sound waves to see,
0:16:52 like, for example, doing like arm raises is different than doing bicep curls.
0:16:52 And those show up.
0:16:58 Yeah. And so we were solving this problem for some populations
0:17:03 that aren’t typically addressed by normal trackers like the Fitbit or something like that.
0:17:08 And so, you know, obese, disabled populations are often not able to use a Fitbit
0:17:13 for exercise tracking because they’re just prescribed different exercises by their doctors.
0:17:17 For example, like I mentioned, the bicep curls with cans or something like that,
0:17:20 or elderly populations as well.
0:17:23 And so we wanted to create this tracker to solve this problem for people
0:17:26 that current solutions just weren’t accounting for.
0:17:29 And yeah, we ended up building this really cool app.
0:17:33 We did a research study where a bunch of us, including myself, went in
0:17:35 and we were doing a bunch of these exercises to collect data.
0:17:37 And that was my first introduction to AI.
0:17:41 We built a classification system that could classify, I think,
0:17:45 around 14 or 15 different exercises at pretty high accuracy, 80 to 85 percent.
0:17:49 And we were also able to count the number of those exercises.
0:17:52 So you get this really holistic tracker where you don’t have to select
0:17:55 the exercise that you’re doing. It’ll write a note. Yeah. Yeah.
0:17:58 So this was your first exposure to AI per se.
0:18:00 That’s such an umbrella term these days, but that’s fine.
0:18:03 Were you intimidated?
0:18:08 Yeah, I really did not know what AI meant or that you could just implement it with,
0:18:11 you know, maybe a hundred, a little over a hundred lines of code, right?
0:18:13 Like that’s that is a crazy concept.
0:18:14 Yeah, yeah, yeah.
0:18:18 System that actually learns with just, you know, by yourself.
0:18:21 And then what was the process of actually, I mean, you said, you know,
0:18:22 one of the things, right?
0:18:24 You’re like, it’s so few lines of code and it’s doing the stuff.
0:18:29 Once you got going, once it was sort of, you know, somebody in the program,
0:18:32 the program itself, whatever it was, kind of showed you how it worked, right?
0:18:33 Took you behind the curtain.
0:18:36 What then? Was it steep learning curve?
0:18:37 Was it sort of natural? Was it?
0:18:41 I’m just kind of curious of the perspective, you know, because when I was in high school,
0:18:43 I took AP Computer Science in high school.
0:18:47 And I think I wrote code on like with a pencil on like one of those exam books
0:18:49 for like my AP final, right?
0:18:52 So different world and imagining like trust my eye.
0:18:54 No, different world for sure.
0:18:56 I think there was a little bit of a steep learning curve for me.
0:19:00 I think that that idea of something, a computer learning was just something
0:19:02 that was so foreign to me, right?
0:19:03 Is yeah, no, it’s still it.
0:19:05 Yeah, yeah, exactly.
0:19:07 It still is. But that’s to you to me.
0:19:10 I think the idea of, you know, how much data we need to train
0:19:13 of a reasonably confident model was was really surprising to me.
0:19:16 You know, we had so many people come in and do so many different repetitions
0:19:20 in these exercises and needed all that for the system to be able to learn.
0:19:23 And so I think that was really an interesting moment for me because I realized,
0:19:25 you know, OK, yeah, I guess that makes sense.
0:19:29 If you have so many different examples of what’s wrong and what’s right,
0:19:33 maybe it makes sense that a computer can learn to identify those patterns
0:19:35 and and predict them in the future.
0:19:37 And so those are sort of some of the initial learnings I had
0:19:39 that sort of helped me wrap my head around this concept.
0:19:42 But I think now, of course, that something from a classification system
0:19:44 to what we’re doing today is is crazy.
0:19:49 Oops. Yeah. So so not to do that time leap here, but for the sake of the episode.
0:19:52 So this project, the project with the exercise tracker,
0:19:55 you were still in high school and undergrad.
0:19:56 Yes, I was in high school.
0:19:59 That was my right after my first experience. Right.
0:20:00 Right after activation. OK.
0:20:03 And so then you studied computer science in college.
0:20:05 Yeah, yeah. I think after all these experiences,
0:20:07 it just really felt like natural path for sure.
0:20:11 Yep. And did you have a focus on machine learning or I mean, what in 60 seconds?
0:20:13 What’s computer science in college like these days?
0:20:17 There’s still a lot of the core education around data structures,
0:20:19 architecture, so I had a lot of exposure to different things.
0:20:24 My honors thesis ended up being around AI and actually an open AI model.
0:20:27 The clip model was what I was working with for image retrieval.
0:20:30 And so that was sort of my focus towards the end of college.
0:20:31 Yeah. Gotcha. OK.
0:20:33 And how long ago did you graduate?
0:20:36 Just last May, so it’s been like a little less than a year.
0:20:37 Yeah. Wow. All right.
0:20:40 So in the workforce, where are you up to now?
0:20:44 Technovation, all these experiences have led to you’re in the thick of it.
0:20:45 You’re working at open AI.
0:20:47 I am working at open AI.
0:20:48 Yes, I’m an engineer on chat.
0:20:50 GPT growth at open AI.
0:20:52 And so it’s a really cool job.
0:20:55 It’s all about our growing, our free and paid user base.
0:20:57 And that ends up being really broad in practice,
0:21:00 but it is a really, really cool job to be at.
0:21:02 I don’t know. It’s so new.
0:21:05 But in a way, I think that makes your your perspective really interesting.
0:21:08 How similar, how different is it now, you know,
0:21:11 doing the stuff, computer science and research and engineering
0:21:15 at a job versus, you know, all of these educational experiences
0:21:16 kind of kind of leading up to it.
0:21:18 Is it is it vastly different?
0:21:22 Does it feel like the next logical step sort of on the on the journey of,
0:21:25 you know, kind of familiar, lots new, but learning and growing?
0:21:27 What’s it what’s it like being out in the workforce?
0:21:30 I think it’s sort of required a mind mindset shift for me.
0:21:33 I think when it comes to college and internships,
0:21:36 there is sort of this concept of some end goal, right?
0:21:38 For so in college, it’s like, let me get through these classes
0:21:40 and I can graduate. I want to learn a lot.
0:21:41 I want to meet my friends.
0:21:43 I want to meet mentors and things like that.
0:21:46 But I think when it comes to working in the industry for me,
0:21:49 the mindset shift was now I’m thinking about what is the impact
0:21:50 that I can have on the world, right?
0:21:52 And how can I maximize that?
0:21:55 And before that, that that was sort of in the back of my head,
0:21:56 but not not really at the front.
0:21:59 And so now it’s sort of what I’m thinking about all day, every day, right?
0:22:01 I don’t have homework and assignments
0:22:05 and all these things going on to to distract me from that from that goal.
0:22:07 And so I think that shift was really interesting.
0:22:12 And that was especially why I was interested in in a team like ChatGBT
0:22:13 growth at OpenAI.
0:22:17 I think one of the things that is really important to me is making sure
0:22:21 that it’s not just people in Silicon Valley using ChatGBT to help their daily lives.
0:22:25 It’s people in, you know, corners of the world where maybe they don’t have
0:22:29 a super powerful mobile phone that can even support an internet connection
0:22:33 for ChatGBT, but maybe they can message ChatGBT through WhatsApp
0:22:37 and get the messages that and and get help and information
0:22:38 and whatever it is that can help them in that way.
0:22:42 And so I think that that sort of shift in mindset from from college,
0:22:45 from learning to working has really helped me shape my goals
0:22:48 and and just feel passionate about about our mission
0:22:50 and what we’re doing with with the product like ChatGBT.
0:22:51 That’s amazing.
0:22:54 I mean, it’s not it’s not my place to say if this makes sense,
0:22:56 but that’s an asset to the world to have that mindset.
0:22:58 So that’s fantastic.
0:23:03 What advice would you give to girls in high school and college
0:23:07 who, you know, are just into the field or interested in computer science
0:23:10 or maybe, you know, the teenager who’s thinking about it
0:23:13 and getting their first exposure to to working with some of the tools?
0:23:15 Oh, this is kind of cool. Where can I go with this?
0:23:18 Any advice? Any maybe learnings from your journey?
0:23:21 Yeah, I think for my journey, I think one of the things I can say
0:23:24 is building confidence early and not being scared of failure.
0:23:28 I think I probably never would have discovered AI if I hadn’t had
0:23:31 I guess the audacity to go after that first research in a new chip
0:23:33 and say, hey, maybe this is a room I belong in.
0:23:36 And, you know, that was not something I felt when I was applying.
0:23:40 I was talking about this boggle program that I built in my computer science class.
0:23:44 I was like, how am I about to jump from this to searching in a lab
0:23:45 and building things for real people?
0:23:49 But I think having that courage to go out and say, I can do this.
0:23:52 That this is an opportunity that is for me as much as it is for anyone else
0:23:55 is something that’s really important to help you to find your path.
0:23:58 You open doors that you never knew were even there.
0:24:01 I think along with that, what goes into building confidence
0:24:05 is finding mentors and a community that can support you in in these things.
0:24:08 And so for me, that came through Technovation, right?
0:24:12 Even now in the Bay Area when I’m back, that community is still there supporting me.
0:24:17 And, you know, just meeting amazing other alumni in the Bay Area was such a surprise for me.
0:24:21 I didn’t know that it would last, that the community would last long for me.
0:24:23 Communities, women, engineers in college, right?
0:24:26 Meeting meeting other people that are doing similar things to you
0:24:28 is the most inspirational thing that you can find, right?
0:24:30 It’s right. If they can do it, I can do it too.
0:24:35 Yeah. And mentors, of course, as well, so many helpful mentors along my journey
0:24:38 had helped me figure out what’s right for me and, you know, wrap my head around
0:24:40 moving from college, going into the industry.
0:24:43 And I think that’s really important to to have some people that have been
0:24:46 through the same things that you’re going through and getting advice
0:24:50 or just even just talking through what you’re going through.
0:24:52 Yeah, absolutely. What is Wiser AI?
0:24:57 Yeah, so Wiser AI is an organization, loosely, that I’ve recently been working on.
0:24:59 It’s it’s more of an initiative.
0:25:02 Just I think what I noticed when I first started working in AI,
0:25:07 even in research and college was we have done so much work to sort of uplift
0:25:08 women in technology.
0:25:12 And there was a big focus, I think, on on that gap when I was in high school
0:25:13 and even middle school.
0:25:15 And I was really seeing feeling that around me.
0:25:20 But as I got into AI and research, I was noticing these are all really,
0:25:21 really male dominated rooms.
0:25:25 And for some reason, I just wasn’t finding as many people talking about
0:25:28 women in AI and underrepresented folks in AI.
0:25:33 And when I first landed the opportunity at Open AI, there was a lot of chatter
0:25:36 about, you know, how quickly AI is moving, you know, the exponential curve
0:25:40 at which these models are performing on these really, really powerful benchmarks.
0:25:44 And that really got me thinking, if we’re building this really powerful
0:25:48 technology that is meant to benefit everyone, how are we going to build that
0:25:52 if we don’t have everyone’s voices involved in developing the system?
0:25:57 And so Wiser AI is an initiative targeted towards that and in bringing
0:26:00 more underrepresented voices in AI, specifically women.
0:26:03 One of the things that I’m really interested in doing with this initiative
0:26:05 is supporting underrepresented founders in AI.
0:26:09 I think that’s a really interesting space where the progress in terms
0:26:13 of like female founders and, you know, the percentage of females founders
0:26:18 supported by investments or even just the percentage of like all women founded
0:26:20 companies is really, really surprisingly low.
0:26:23 I think that was an interest of mine at one point, right, becoming a founder.
0:26:26 And it’s it’s really surprising to see that that’s still.
0:26:32 Yeah, it’s it’s so unreflective of the actual, you know, population
0:26:34 of people walking around the planet. It’s yeah.
0:26:35 And so you started Wiser.
0:26:37 Yes. Yeah. Yeah.
0:26:38 Bringing together a team right now.
0:26:41 I think one of the really big things that we want to do this year and sort
0:26:45 of the big target for the organization this year is to actually hold a
0:26:49 conference, bringing together underrepresented founders in AI, seeking
0:26:52 people that can talk about what it means to build responsible AI and how
0:26:54 we can incorporate that into our systems.
0:26:58 And yeah, just just helping bring together a community that can support each other.
0:27:03 Amazing. So, Tara, I want to sort of welcome you back for this kind
0:27:06 of last sort of wrap up thing and ask both of you.
0:27:09 And I feel like this is one of those questions that like shouldn’t need
0:27:13 to be asked out loud and answered because it should be obvious, but it’s not.
0:27:16 Why is is the work you’re doing so important?
0:27:21 And why is it so important to work for equitable and inclusive AI right now?
0:27:25 You’ve talked about it throughout the episode, but to kind of put a point
0:27:28 on it and really bring it to bear, why is this so important right now?
0:27:33 As I was saying, like the large language model has been built on the majority,
0:27:37 right, like the majority language on the Internet, which may not
0:27:39 necessarily reflect the reality.
0:27:44 And so you actually get boring results because it’s the generic
0:27:46 very, very the lowest baseline, right?
0:27:52 And so what’s the fastest way to make more innovative products, right?
0:27:53 Is to open it up, right?
0:27:55 And I think there could be many ways of doing it.
0:28:00 I think one obvious way is look, 70 percent of the developers
0:28:04 are of a particular background and from a particular cognitive perspective,
0:28:08 right? And so when you have different people around the room, I mean,
0:28:12 it just, it came from India last night, and it’s fascinating, right?
0:28:15 Like just going to a different place and understanding how people
0:28:19 in different cultures think it just increases your sort of scope, right?
0:28:21 So it’s just basic stuff.
0:28:26 So I’m not big on business axioms per se, but isn’t there just a basic guideline
0:28:29 of sort of product development that, you know, the more that what you’re developing,
0:28:32 the more that people building the thing sort of reflect the people who are going
0:28:36 to be using it, the better off you’re likely, you know, the thing you’re
0:28:37 making is likely to be now.
0:28:41 Absolutely. And the market is beyond the U.S., right?
0:28:44 No, I mean, this is, this is everything you’re building for the world.
0:28:48 So to me, it’s just a very basic common sense argument that you want
0:28:51 a diverse group of people innovating.
0:28:57 And I will say that the challenge is the big one because women represent
0:29:00 I think the largest group of people that have been historically
0:29:03 discriminated against as a whole, right?
0:29:07 So I think 151 countries legally discriminate against women.
0:29:10 And so social norms are very deeply rooted, right?
0:29:13 And I think men and women are extremely different.
0:29:17 So I think that the conversation can be very nuanced and subtle.
0:29:22 But from an economic point of view, I think you are missing out if 50%
0:29:27 of the population isn’t represented in your software development teams.
0:29:30 You know, it’s not enough to say we have equal representation
0:29:32 in sales or HR or whatever.
0:29:34 I think it has to be in the software development team.
0:29:39 So I think everybody will benefit when the products are less boring.
0:29:44 Yep. Anshita, in many ways, this whole conversation, particularly you and I
0:29:47 speaking, you talking about your work in the past several minutes, you know,
0:29:51 address this question, but specific to talking about the importance of working
0:29:54 for Equitable Inclusive AI right now, anything you’d like to add?
0:30:00 Yeah, I think overall we’ve already seen the effects of building bias systems.
0:30:03 In systems like predictive policing or in health care.
0:30:07 We know there’s been many studies about how this affects
0:30:09 disproportionately affects underrepresented populations.
0:30:14 I think at the rate at which AI is developing right now, these systems
0:30:18 are going to be everywhere, not just in health care, but also in finance,
0:30:19 hiring everywhere.
0:30:23 And so I think it’s just we’re at a crucial point where if we don’t
0:30:26 start doing that work towards making these systems inclusive, when they’re
0:30:28 everywhere, it’s hard to take that back.
0:30:30 I think it needs to happen now.
0:30:35 For listeners who want to learn more, want to get involved,
0:30:39 maybe want to see if they can join Technovation, any and all the above,
0:30:44 all the great work you’re both doing, where can folks go online to learn more?
0:30:47 Again, I’ll plug Episode 85, Tara, your first appearance on the pod.
0:30:51 But for more information about Technovation, Tara, where can people look?
0:30:54 Yeah, I would say the Technovation season is ongoing right now and you’re
0:30:59 actively looking for mentors, so and you don’t need to have a technical
0:31:02 background to be a mentor, because what you need to bring to the table is
0:31:07 just this courage to say, I don’t know, let’s go find out together.
0:31:08 Because guess what?
0:31:11 You probably don’t know because these things are changing so fast, but it’s
0:31:15 an incredible experience where you’re project managing and cheerleading
0:31:18 a team of girls to actually build an AI prototype within three months.
0:31:23 So it’s an amazing way to actually go through an accelerator yourself.
0:31:27 We’ve had stories where mentors have applied to my combinator and gotten in
0:31:30 after they went through Technovation because Technovation gave the mentors
0:31:32 the courage to start their own business.
0:31:36 So I’d say 95 percent of the time when girls match with mentors,
0:31:37 they finish the full program.
0:31:41 So mentors are the one of the most important pillars for our program.
0:31:43 So please sign up at Technovation.org.
0:31:45 Technovation.org?
0:31:45 Yeah.
0:31:46 OK, great.
0:31:48 And Chita, is there wiser?
0:31:50 Can people find out more about wiser online?
0:31:54 Yes, and I’m actively looking for women to work with to organize this
0:31:56 conference that we have a goal of having this year.
0:31:57 Do you have a location in mind?
0:32:00 It might have to be around the Bay Area just because of production,
0:32:04 like maybe in the Bay Area, but it’s still still working on a location.
0:32:06 Honestly, so open to any suggestions.
0:32:10 But we would love to have more talented, amazing, passionate women to work with.
0:32:16 They could reach out directly or wiser.ai.org is the website.
0:32:17 Excellent.
0:32:19 Well, again, thank you both so much, Chita and Chita,
0:32:23 for taking the time to come on the podcast and share the work you’re doing.
0:32:28 It’s just really great, inspiring work to learn about and for people to get involved with.
0:32:32 And, you know, anybody out there who’s considering maybe being a mentor,
0:32:34 if you’ve never had that experience, I haven’t worked with Technovation,
0:32:38 but I’ve done other related things and it’s the best.
0:32:41 And not only that, but you get to go through an accelerator yourself.
0:32:42 It’s kind of hard to beat, right?
0:32:43 Let’s do it again.
0:32:44 Let’s catch up down the line.
0:32:48 But it was an absolute pleasure to meet you and Chita and all the best to both of you
0:32:49 and the work you’re doing.
0:32:50 Thank you, Noah.
0:32:51 This was a pleasure.
0:32:52 Yeah, thank you so much, Noah.
0:32:53 And thank you, Tara.
0:32:56 It’s just so inspiring to hear about the work that you’re continuing to do
0:32:58 and knowing how much of an impact it’s already had on me.
0:32:59 So thank you.
0:33:00 That made my day, right?
0:33:03 Like, listening to alumna and saying, OK, all the hard work
0:33:05 and your vision of improving the world.
0:33:07 Like, that’s what we work every day for, right?
0:33:11 So I wouldn’t have the energy to do what we do in the battles to fight
0:33:13 if it weren’t for the alumna, so it’s mutual.
0:33:16 [MUSIC PLAYING]
0:33:17 .
0:33:18 .
0:33:20 .
0:33:24 [MUSIC PLAYING]
0:33:27 [MUSIC PLAYING]
0:33:31 [MUSIC PLAYING]
0:33:34 [MUSIC PLAYING]
0:33:38 [MUSIC PLAYING]
0:33:41 [MUSIC PLAYING]
0:33:45 [MUSIC PLAYING]
0:33:48 [MUSIC PLAYING]
0:33:52 [MUSIC PLAYING]
0:33:55 [MUSIC PLAYING]
0:33:58 [MUSIC PLAYING]
0:34:02 [MUSIC PLAYING]
0:34:12 [BLANK_AUDIO]

0:00:10 [MUSIC]
0:00:13 >> Hello, and welcome to the NVIDIA AI podcast.
0:00:15 I’m your host, Noah Kravitz.
0:00:19 Our guests today were recently featured on the NVIDIA blog for their work in
0:00:22 creating Ulangizi, an AI chatbot that delivers
0:00:25 multilingual support to African farmers via WhatsApp.
0:00:28 As vital a project as it is, however,
0:00:31 GUI AI is much more than a single chatbot.
0:00:33 GUI AI is a platform for developing
0:00:37 low-code workflows built on private and open source AI models.
0:00:41 Combining ease of use with innovative features like golden Q&As,
0:00:44 GUI enables developers to code fast and change the world.
0:00:48 Here to tell us the GUI story are the company’s founder and CEO,
0:00:52 Sean Blagsvet, and founder and chief creative officer, Archana Purcell.
0:00:56 Welcome to you both and thanks so much for joining the NVIDIA AI podcast.
0:00:57 >> Hello.
0:00:59 >> Hi, thanks, Noah.
0:01:02 >> So there’s a lot that I’m looking forward to you getting
0:01:05 into about the GUI platform, how it started,
0:01:06 all the things it can do,
0:01:11 including how you’re helping developers combat AI hallucinations,
0:01:12 which is a big topic these days.
0:01:15 But I’d love it if you can start at the beginning and
0:01:18 tell us what GUI AI is and how you got started.
0:01:20 >> Let me take a shot at that.
0:01:24 We could start it from actually a digital arts project funded by
0:01:26 the British Council many moons ago,
0:01:29 I’d like to say 2018, 2019,
0:01:33 where we applied to create an AI persona that would match,
0:01:36 make creators, activists,
0:01:39 designers from across borders of the UK and India.
0:01:43 We won that award, we built out a prototype,
0:01:45 we tested it, it worked beautifully,
0:01:47 and long story short,
0:01:52 we managed to get a seed fund from Techstars in the race.
0:01:55 >> We got into Techstars, which was an excellent program.
0:01:58 We took this idea of an AI persona and built
0:02:02 an entire communications app around it called Dara.network,
0:02:04 meant to service cultural organizations and
0:02:06 social impact organizations,
0:02:10 enable them to manage their alumni and keep in touch with each other easily.
0:02:14 The first AI persona we built also confusingly called Dara,
0:02:18 feeling lonely and wanted some friends and we thought,
0:02:22 wouldn’t it be great to invite non-tech folks,
0:02:24 writers, playwrights, all those poets,
0:02:26 could we have them come in and craft
0:02:29 their own AI personas from scratch?
0:02:31 >> This is right in the middle of COVID.
0:02:33 So all those are out of work, right?
0:02:35 >> They’re isolated.
0:02:36 >> Isolated, yep.
0:02:42 >> Yeah. So we invited 23 folks from across the UK,
0:02:45 the US, India, Sri Lanka even,
0:02:48 and met constantly literally every week,
0:02:50 and ended up developing pretty much
0:02:54 an underlying architecture that enable them to build out,
0:02:55 what one might call it.
0:02:57 >> Hiring task-tasking video box, right?
0:03:04 So our co-founder Dave was hanging out in Discord forums with,
0:03:06 what’s his name, Brockman,
0:03:08 with the president of OpenAI until I think it was-
0:03:09 >> Brockman, yeah, yeah.
0:03:11 >> Yeah, it’s like five years ago,
0:03:15 and then so we had really early access to the GPT APIs,
0:03:17 and we had already built Dara
0:03:20 as an asynchronous video messaging platform,
0:03:22 kind of like Discord plus LinkedIn
0:03:25 with a little bit of Mark Polo in there.
0:03:27 And so the thought was, well,
0:03:29 what if we took, it was kind of a wild idea,
0:03:32 like what if we took the video messages that people sent,
0:03:35 pulled that up with Google’s speech recognition,
0:03:39 fed that to a long-form script, right?
0:03:42 These playwrights and authors were putting together,
0:03:43 and then we had this thought of like,
0:03:47 what is it if it’s $1.50 to $2 per API call?
0:03:48 What could we get, right?
0:03:50 And then so we basically had this script
0:03:53 that they were crafting and they were writing together.
0:03:57 And then we had the deepfakes APIs
0:04:00 were just beginning to come out and there was text speech.
0:04:01 So we’re like, well,
0:04:03 we can take what the bot says back, right,
0:04:07 that DaVinci output and then take that little text,
0:04:09 put it to a text-to-speech engine,
0:04:11 put it into a lip sync piece,
0:04:15 and then, boom, you’ve got these Turing test passing characters.
0:04:16 We called them the Radbots.
0:04:17 And those were the Radbots.
0:04:18 They were awesome.
0:04:20 They still are kind of awesome.
0:04:22 Yeah, they were crazy.
0:04:23 And we had a little thing,
0:04:25 usually this is a public family podcast,
0:04:28 but we had the Radbots say.
0:04:32 That was the Wild West of LLMs back in 1921, yeah.
0:04:35 Yeah, and these bots really spoke their mind.
0:04:38 And they just spoke their mind.
0:04:42 They represented issues that the writers brought in,
0:04:44 that they felt were not represented well enough,
0:04:48 spoke on behalf of communities that were underrepresented,
0:04:51 kind of in a bid to reduce the algorbias
0:04:54 that the group was feeling quite intently.
0:04:56 You know, we are in, the bots are done,
0:04:58 they’re happily chatting away, people are excited.
0:05:01 And like little kids have talked to them a thousand times,
0:05:03 like one kid, 1200 times, right?
0:05:06 And these were presented at, you know, the…
0:05:09 India Art Fair, the, you know,
0:05:12 Be Fantastic, the British Council hosted the event,
0:05:14 the India Literature Festival.
0:05:19 They went a little bit out, quite a bit.
0:05:21 The point was that we had built out
0:05:23 this underlying architecture and orchestration platform
0:05:27 pretty much that enabled us to kind of plug and play
0:05:29 all the incumbent new technologies
0:05:31 that were emerging in AI.
0:05:35 And that was a moment, I think, in this very room,
0:05:37 actually, when John and I were like,
0:05:40 “Okay, we have the messaging platform,
0:05:41 “but what can we do more?”
0:05:44 And that was pretty much the start of GUI.
0:05:47 We felt, hey, if we can take artists and writers
0:05:49 with us on a journey, why not open it up
0:05:52 into a wider world and kind of really take this,
0:05:55 you know, this mission we had begun to feel quite deeply
0:05:57 by then of like democratizing AI
0:06:00 and really allowing people to play with it,
0:06:03 even if they didn’t know how to code, I’m not a coder.
0:06:04 So, you know, I feel that part.
0:06:06 I want to ask you quickly about your backgrounds.
0:06:08 I was wondering kind of both technically
0:06:12 and at the risk of making, you know, being corny,
0:06:15 aside from the fact that enabling everyone,
0:06:19 artists, creatives, non-coders to use the technologies,
0:06:21 aside from the fact that that’s just an awesome thing to do,
0:06:23 why was that your focus?
0:06:25 How would you do such a thing?
0:06:28 Well, my background is actually in art and design.
0:06:31 I studied painting and animation film design
0:06:34 back in the day, and quite wonderfully met Sean.
0:06:39 He had landed into India to set up the Microsoft Research Lab.
0:06:41 This was in 2005.
0:06:43 So, I was number three and she was number five.
0:06:45 In that amazing organization.
0:06:47 We’re going to track down number four, get them on the pie.
0:06:49 Yeah, yeah, yeah.
0:06:52 Yeah, so we, that’s how we met.
0:06:54 And I don’t have a coding background,
0:06:56 but I did get to hang out at Microsoft Research,
0:06:58 get access to some of the brightest minds in the world
0:07:00 and the papers that were coming out of there.
0:07:03 And we were the, I’d say the first
0:07:06 and maybe the only project, what is it?
0:07:07 Advanced prototyping.
0:07:09 My title was Head of Product Management
0:07:11 and Advanced Prototyping.
0:07:13 And so, we did a lot of work.
0:07:17 You know, this is like 2004 to 2010, around,
0:07:19 you know, how do we build interfaces that reach everybody?
0:07:23 How do we build interfaces for folks where literacy is an issue?
0:07:24 You know, I have the patent on, you know,
0:07:28 machine translation and instant messaging from 2005.
0:07:31 This is a space that we have been in for a long time,
0:07:33 running big models on NVIDIA hardware,
0:07:36 trying to understand what everybody has to say
0:07:37 through channel interfaces.
0:07:41 And frankly, India was a head that skipped email, right?
0:07:43 And everybody kind of went directly
0:07:46 into SMS-based interfaces into WhatsApp in the years ahead.
0:07:50 So this is a space that I would say we’ve been experimenting
0:07:52 in literally for 20 years in terms of like,
0:07:57 how do you build tools and interactive AI-based interfaces
0:07:58 that work for everybody, including people
0:08:01 that don’t read or write in English very well.
0:08:04 And then after that, Archana went off and…
0:08:05 – Set up an organization called JAGA
0:08:08 and then another called Be Fantastic,
0:08:10 with co-founders Freeman and Kamiya.
0:08:14 And the idea was really, how do we take artists
0:08:17 and creative practice and technology practice?
0:08:18 How can we bring practitioners
0:08:20 from these different fields together
0:08:23 and enable them to have conversations
0:08:26 and service kind of some of the big pressing issues
0:08:27 of the time.
0:08:30 So soon a lot of activists and all that started.
0:08:33 We started JAGA in 2009, Freeman and I,
0:08:38 and Be Fantastic kicked off as a public arts festival
0:08:39 looking at arts and technology,
0:08:44 particularly towards climate change and UNSDG kind of issues.
0:08:45 – And then meanwhile, I was off running
0:08:47 a company called BalaJob,
0:08:49 which was the largest informal sector-focused job site
0:08:50 in India.
0:08:53 So how do we take drivers and cooks and maids
0:08:55 and basically using the phone and IVR.
0:08:58 This is an SMS and multi-lingual interfaces
0:09:00 basically hook them up with better jobs.
0:09:03 – Pretty deep phone interface as a way.
0:09:04 – Yes, she was the voice in,
0:09:07 but we did 50,000 phone calls a day
0:09:09 or telephone bills were astronomical.
0:09:11 We had nine million users reprocessed
0:09:13 and a million applications a month.
0:09:15 That was a big initiative that was 11 years of my life
0:09:18 while she was running JAGA in parallel.
0:09:19 And over there we got there.
0:09:20 – Yep.
0:09:23 So you’ve got the credentials, you’ve got the chops.
0:09:26 And I think before I interrupted you,
0:09:27 Archana, you were about to say,
0:09:30 and so we wound up with this platform that we called GUI.
0:09:31 – Yes.
0:09:32 – We did, and I love the name.
0:09:35 It’s a little take on, you know, graphic user interfaces.
0:09:35 – Yep, yep.
0:09:38 – We talked about it all the time when GUI was a thing.
0:09:40 – Also the connective tissue, right?
0:09:42 – And the GUI love, you know.
0:09:46 Yeah, so we kind of pivoted pretty much overnight,
0:09:49 quite literally, we took our team over to the idea.
0:09:51 They really liked it as well.
0:09:52 And pretty soon we had GUI,
0:09:54 which is pretty much what it is today.
0:09:55 – Well, and then, you know,
0:09:57 it was sort of founded on this premise
0:09:58 that almost started out as a joke,
0:10:01 but feels less and less like a joke every day,
0:10:04 which is that all of us are just gonna be AI prompt writers
0:10:06 and API stitchers, right?
0:10:07 In terms of like the job of the future
0:10:09 that everyone will do seem to be that.
0:10:11 And then, you know, this thought being,
0:10:15 well, if that’s the world, what kind of things do you need?
0:10:17 Like what would be the JS equivalent,
0:10:18 or JS fiddle equivalent of that?
0:10:21 As in, how, when I make something,
0:10:22 do you get to view source on that
0:10:24 and understand what I’m doing so that I can learn
0:10:26 by building on top of what you’ve done?
0:10:27 – Right.
0:10:29 Like also from the, you know, publishing.
0:10:30 I mean.
0:10:31 – Yeah.
0:10:33 Then research, both of citations and public play papers,
0:10:36 which goes back to the beginnings of the light, right?
0:10:38 And so, and the open source movements as well.
0:10:41 We wanted to say, what would that mean
0:10:44 for these new higher level abstractions
0:10:46 of, hey, you got a little bit of LLM prompts
0:10:49 and you want to get over to hear this other API
0:10:51 and you want to connect to some other communication platforms.
0:10:53 How do you extract that up
0:10:55 to sort of allow a whole new generation
0:10:57 and non coders to basically play?
0:10:59 – And efficiently too, right?
0:11:02 I mean, the idea being that it’s a one-stop spot, right?
0:11:06 You can try all kinds of different AI technologies
0:11:09 and tools without subscribing individually to each other.
0:11:09 – To any one of them.
0:11:11 – Yeah, so it’s, yeah.
0:11:12 – And so that’s because we do this part of,
0:11:15 we saw a huge amount of innovation
0:11:17 coming from many sectors, right?
0:11:20 The open source ecosystem was clearly making
0:11:22 new incredible models every day.
0:11:23 – Right.
0:11:25 – And not just around LLMs, but around animation
0:11:28 and image creation and text speech, right?
0:11:30 And we saw, given our work with Radbots,
0:11:33 that when you allow people who are creative
0:11:36 and empowered to put those together in novel ways,
0:11:38 you get this magic.
0:11:38 – That’s where you get the magic.
0:11:40 – And it’s thousands of interactions,
0:11:42 which is definitely bigger than the sum of its parts.
0:11:43 – Yes.
0:11:46 – And so we wanted very specifically to say, great,
0:11:49 when opening AI or Google or the open source community
0:11:51 comes out with some new feature,
0:11:54 we should constantly allow the ecosystem to get better.
0:11:56 So we, with this whole thought of,
0:11:59 we should abstract on top so that every component
0:12:02 is essentially hotswappable and evaluatable,
0:12:04 which is where your golden questions things come in.
0:12:05 – Yes.
0:12:08 – That you can basically say, hey, open AI 4.0
0:12:11 or, you know, some 01 piece came out,
0:12:13 is it better, cheaper, faster for me?
0:12:15 And then, you know, given our impact piece
0:12:17 of like going forward, like, is it also,
0:12:20 what’s the carbon usage of each one of them, right?
0:12:23 And how do we make that clear to those people
0:12:26 that are buying and using those APIs,
0:12:29 as well as sort of the fourth important factor
0:12:33 for any chain or workflow that you’re gonna put together?
0:12:34 – So many questions.
0:12:36 Maybe I’ll ask you, and this is sort of a bad question
0:12:39 for an audio only format, but we’re gonna,
0:12:39 that’s what we do here.
0:12:40 So we’re gonna do it.
0:12:42 A new user comes to GUI.
0:12:43 – Yes.
0:12:45 – Someone who perhaps, you know, understands
0:12:48 when an LLM is, how technology works.
0:12:49 They know what an API is.
0:12:51 Maybe they copy and pasted some code once or twice.
0:12:54 You know, how do they get started?
0:12:56 Is it drag and drop?
0:12:58 Is it writing things into, you know,
0:13:01 sort of a chatbot, text-only interface?
0:13:03 How does the platform actually work for the user?
0:13:05 – So at a high level, you know,
0:13:07 this is something we took from Bob a Job.
0:13:09 We did pretty good at SEO.
0:13:11 And the reason that we do well at SEO
0:13:13 is we try to lower friction.
0:13:18 So when you come in, if you Google like AI document search
0:13:21 or, you know, agricultural bot or AI animation,
0:13:23 you can come in there and you can see the prompts
0:13:25 and the output directly.
0:13:27 And you can go into examples
0:13:28 and you can see a bunch of other ones
0:13:30 that we hopefully gets better and better
0:13:33 and are relevant for your field is sort of a UGC model.
0:13:34 And every single one of them,
0:13:35 you can just say, great, I like that.
0:13:36 I’m gonna tweak it, right?
0:13:38 – Right, so you grab a pre-exhausting.
0:13:40 – Yep, I’m gonna change what this model is.
0:13:43 I’m gonna, and so we’re always having this kind of,
0:13:45 we’re definitely inspired by like Replicate, right?
0:13:47 Which is definitely, you know, this idea of like,
0:13:49 what are the inputs somebody else used?
0:13:50 What are the outputs?
0:13:52 But to do so in a way that we’re chaining together
0:13:54 many of these different components,
0:13:56 just basically see something also.
0:13:57 So that’s kind of it.
0:13:58 – And it’s going to be a drag and dropy
0:13:59 and more kind of pull-down menu.
0:14:01 – Yes, yeah.
0:14:03 – ‘Cause the idea there is transparency.
0:14:04 – Yeah.
0:14:05 – Like for a lot of other sites,
0:14:08 I would argue they hide the prompt
0:14:10 that’s actually making the magic happen.
0:14:12 Or they hide the model that’s making the magic happen.
0:14:15 And so for us, we have a belief of like,
0:14:19 no, the model that you’re using for today versus tomorrow
0:14:21 and all of the prompts and everything else.
0:14:22 – Small change of the black box here.
0:14:25 It’s all digestible and viewable and respectable
0:14:26 all the way down.
0:14:29 – We kind of call these recipes for long times.
0:14:30 – Yeah, sure.
0:14:31 – Before we close it.
0:14:32 – Yes.
0:14:32 – Really like, these are the ingredients.
0:14:35 This is how we made this mix and, you know,
0:14:37 you can make your starting from there, yeah.
0:14:38 – So everything is forkable.
0:14:40 And then again, just like JS Fiddle,
0:14:42 you make one change, that’s a new URL.
0:14:43 You can share it with your friends.
0:14:45 You know, they literally next week
0:14:46 will come out with workspaces.
0:14:48 So you can work on these things collaboratively
0:14:49 with version histories.
0:14:52 So you can say, hey, I have a static endpoint
0:14:55 of like my cool co-pilot, we can work on that together.
0:14:57 And then you can do things like hook that up
0:14:58 directly inside the platform
0:15:00 to things like WhatsApp or Slack.
0:15:01 – Facebook.
0:15:02 – Or Facebook.
0:15:04 And that’s actually, I feel like an underestimated part
0:15:08 of getting these things to work on the communication tools
0:15:12 that really help you is way harder than you have.
0:15:14 – Well, so I wanted to–
0:15:16 – I have a friction there that we try to take away.
0:15:18 – Right, and I wanted to mention, you know,
0:15:20 and I don’t know, this is back in the day
0:15:22 I covered the mobile phone industry.
0:15:25 And I don’t know, maybe we have a great audience
0:15:26 so they probably know, but, you know,
0:15:29 for sort of a US centric point of view,
0:15:31 people don’t necessarily understand
0:15:33 that in so many parts of the world,
0:15:35 your phone is your computer period.
0:15:37 And people are sharing phones or, you know,
0:15:39 getting a phone to use for a day or that kind of thing.
0:15:40 But it all happens on the phone.
0:15:43 There’s no laptop, there’s no desktop workstation,
0:15:44 all that stuff.
0:15:46 And so when I was, you know, researching,
0:15:50 prepping for the taping and reading about how,
0:15:53 Oolongeezy, the farmer chat bot went through WhatsApp.
0:15:55 You know, it was like, oh, cool.
0:15:56 And I was like, well, of course it does
0:15:58 because that’s how people work.
0:16:01 So, you know, maybe to get back to what you were saying,
0:16:03 Sean, about putting these tools
0:16:05 into the communication platforms.
0:16:07 What were some of the hurdles, some of the challenges?
0:16:09 Maybe some of the pleasant surprises and working on that.
0:16:11 Oh, there’s a ton, right?
0:16:14 And so we’ve got a talk that’s on our site
0:16:17 called like the stuff that goes wrong, right?
0:16:19 Which is basically like, so, you know,
0:16:22 right after we began, again, our old friend
0:16:23 and, you know, director of the lab,
0:16:28 Annandana said, hey, Rickon who runs Digital Green
0:16:31 is working and the whole space is like, you know,
0:16:34 bots for development is going to be a thing.
0:16:36 Which is basically getting this part of,
0:16:38 if we want to convince every farmer in the world
0:16:41 to basically change their livelihood
0:16:44 and the crops they grow because climate change
0:16:46 is necessitating that over the next decade.
0:16:48 We’ve got to convince all of them
0:16:51 to change literally the source of their income.
0:16:53 That is a hard challenge.
0:16:55 Every government in the world has this challenge
0:16:57 for the, I don’t know, several billion people
0:16:58 on earth that are farmers.
0:17:01 So there’s this piece that he recognized like,
0:17:03 okay, bots are going to be a thing.
0:17:04 Why don’t you get together?
0:17:07 And so what we did there was to say, hey, you know,
0:17:09 in the case of Digital Green,
0:17:12 they had an incredible library of thousands of videos,
0:17:16 basically of one farmer recording how they do a technique
0:17:19 better that was then shown to other nearby farmers.
0:17:20 – Farming best practice.
0:17:21 – Farming best practice.
0:17:24 There’s also, you can think of it as like all of the fact
0:17:26 of every question that everybody’s asked
0:17:28 and the ag space that goes to the government
0:17:31 and then a bunch of like local knowledge
0:17:33 in the form of like Google Docs of like,
0:17:36 what people should do coming from local
0:17:37 on the ground NGOs.
0:17:39 And so what we did is to say, hey,
0:17:41 we’ve built this extensive platform.
0:17:42 We can have rad bots.
0:17:43 We know how to do speech recognition.
0:17:47 Well, we’re running private keys to all of the best services.
0:17:50 Plus we have our own a 100 infrastructure
0:17:51 and GPU or core orchestration.
0:17:53 So we can run any public model too.
0:17:54 So then we can say, great,
0:17:56 we can take all those videos which are not in English,
0:17:59 right, transcribe them, basically use a bunch
0:18:02 of DP24 scripts to create synthetic data around them
0:18:04 so that it’s not just this transcript,
0:18:07 but it’s also like, what is the question
0:18:09 that a practitioner might actually ask
0:18:10 and what’s the answer here?
0:18:13 And then use all of that to basically shove that
0:18:15 into a big vector DB, right?
0:18:18 And then say, okay, we then hook that up on WhatsApp.
0:18:20 And then you put in translation APIs
0:18:22 and speech recognition APIs in front of that.
0:18:24 And then boom, you suddenly have something
0:18:28 that works in multiple languages in multiple countries
0:18:31 using locally referenced content with citations back
0:18:34 that can speak any language that is actually useful
0:18:36 to folks on the ground or small shareholder farmers.
0:18:39 – That was what we demoed at the UN with Rican
0:18:43 in April, 2023 at their general assembly science panel, right?
0:18:46 And so you now look across the world,
0:18:47 do you have bots or a thing?
0:18:49 And I’m not saying like, obviously we weren’t,
0:18:51 the only people involved in this kind of transition.
0:18:54 But the thing that I think for us was exciting
0:18:58 is a bunch of people in the private sector also noticed.
0:19:00 And they said, hey, if you’re looking
0:19:03 at how do I make frontline workers productive,
0:19:06 people that need to fix your AC or do plumbing.
0:19:06 – Right, right.
0:19:08 – They have the same issues of like,
0:19:11 I need to aggregate every manual
0:19:13 of every AC sold in America,
0:19:15 plus all of the training videos around them,
0:19:18 plus ask any hard question in order for me to do my job.
0:19:20 And oh, by the way, all the master technicians
0:19:22 in that field retired with COVID, right?
0:19:24 And so there’s none left, right?
0:19:27 And so, but the technology that you need
0:19:29 to make that happen is actually the same.
0:19:32 And hence, you know, you’ll see us,
0:19:34 we talk a lot about frontline worker productivity
0:19:36 because I think we do this really well
0:19:40 by essentially aggregating all of these different parts.
0:19:41 That was the long answer.
0:19:44 – Yeah, one of the things that you mentioned a few times
0:19:48 is languages and, you know, a lot of the models,
0:19:50 I mean, English for better or for worse
0:19:54 is taking over, spreading, ubiquitous, et cetera, right?
0:19:56 And a lot of the models trained on English,
0:19:58 you’re working with all kinds of languages,
0:20:00 including, from my understanding,
0:20:03 tons of local dialects and, you know,
0:20:05 the kinds of things that the models
0:20:07 aren’t necessarily trained on.
0:20:09 Tackling that, right?
0:20:12 Talking about translations and all that kind of stuff.
0:20:15 Are you also working with, you know,
0:20:18 training foundational models in these languages,
0:20:22 or is it just a better way to tackle it by doing,
0:20:24 and I may have this wrong, so please correct me,
0:20:25 but doing what I think I understood
0:20:27 as translating back to English
0:20:30 and then using that to work with the LLMs.
0:20:33 – Again, it goes back to the sort of core philosophy of GUI
0:20:35 that we always wanna be the super set
0:20:37 of everything else out there.
0:20:41 I personally think as a small startup by small,
0:20:43 I mean, under a billion dollars in funding,
0:20:45 it is fool’s errand to try to train
0:20:48 any foundational models, right?
0:20:50 Because every six months you’re gonna be outclassed.
0:20:53 And so I’m gonna leave that to the people
0:20:55 that can put a hundred billion or more into it.
0:20:58 And yet, every single day I wanna know,
0:21:01 does that work better for my use case?
0:21:04 And we take this very use case specific
0:21:06 evaluation methodology, which is this golden questions,
0:21:10 and then apply that to, hey, I have 50 farmers
0:21:12 outside of Patna in India,
0:21:16 speaking this particular dialect of Bhojpuri, right?
0:21:18 Here’s the questions that they ask.
0:21:20 Here is the expert translation or transcription
0:21:21 into Bhojpuri.
0:21:23 Here’s the expert translation of that question.
0:21:25 That is my golden set.
0:21:28 And then what we allow you to do is to say,
0:21:31 I’m gonna run this essentially custom made
0:21:34 evaluation framework across every model
0:21:36 and every combination of those things,
0:21:39 so that this week I can tell you, huh,
0:21:42 the Facebook MMS large model works actually better
0:21:45 than Google’s USM, which may suddenly work better
0:21:48 than GPT4O audio, right?
0:21:52 And to basically allow organizations to evaluate
0:21:55 which of the current state of the art models,
0:21:57 and in particular, the combinations of those
0:22:00 work best for their use case.
0:22:02 So we have an evaluation level, not the training level.
0:22:05 – Right, is that a hands-on user thing,
0:22:08 figuring out which model, which combinations to use,
0:22:11 or is that something the platform does for the users?
0:22:12 – That itself is another workflow.
0:22:14 So goo.ic/bulk, right?
0:22:16 You can upload your own golden data set,
0:22:19 and then you can then say, great, I wanna do us,
0:22:21 and again, you can see all of the work
0:22:23 that we’ve done for other organizations,
0:22:25 and then you can just sort of say, great,
0:22:28 this is how they did it, I can copy, not copy,
0:22:30 I can just fork their recipe on the website.
0:22:32 – And the advantage there is you don’t have to run
0:22:35 the DevOps to run all of those new state of the art models.
0:22:36 – Yeah, absolutely.
0:22:40 I’m speaking with Sean Blagsfett and Archna Prasad.
0:22:43 They are the co-founders of GUI AI,
0:22:48 a low code change the world, literally change the world.
0:22:51 A lot of people say that, but I think y’all are doing it.
0:22:54 Change the world platform for using AI models
0:22:56 or all kinds of things, but we’re talking particularly
0:22:59 about frontline workers, be it an HVAC technician
0:23:03 or a farmer in a rural community in Africa.
0:23:06 Sean, you mentioned, I tease this at the beginning,
0:23:08 you talked a little bit now about the golden sets
0:23:11 and the golden Q&As, so I wanna ask you about that
0:23:13 and about issues around hallucinations.
0:23:17 It’s one thing if I’m using a chatbot to help me
0:23:19 in my writing work, and it hallucinates,
0:23:20 and I can sort of read it.
0:23:23 It’s another thing if a founder or anybody else
0:23:26 is asking a chatbot for best practices
0:23:29 for their livelihood, hallucinations literally,
0:23:31 life or death there, how do you deal with that?
0:23:35 – So there’s a variety of techniques that I’d say out there.
0:23:37 You should be suspicious of anytime anybody says
0:23:40 we’re 100% hallucination free in general.
0:23:43 So there’s the rag pattern, which says,
0:23:46 hey, I will search your documents or video
0:23:49 or whatever you put in there and I’ll only return,
0:23:51 well, then you get back those snippets
0:23:53 and then you ask the LM to summarize it.
0:23:56 The risk of hallucination there goes down, right?
0:23:58 Because you said, hey, I’m summarizing
0:24:00 some simple paragraphs.
0:24:04 That’s probably okay, honestly for things like farming.
0:24:07 It may not be okay for things like healthcare
0:24:09 because the other thing that happens often
0:24:11 in our pipelines is you take that, you know,
0:24:14 kind of summarization and then you do a translation.
0:24:17 And that translation, you know, for English to Spanish,
0:24:19 great, we’re not gonna probably have a problem,
0:24:22 but English to Swahili, English to Kikwa,
0:24:23 you’re like, I don’t trust that.
0:24:27 So without other techniques that we see out there
0:24:30 where if you really wanna do hallucination free,
0:24:34 then what you do is you sort of translate the user’s query
0:24:36 into a vector search of which question
0:24:38 that’s already in your data bank
0:24:41 whose answer has already been approved by say a doctor.
0:24:44 Does your question most align to?
0:24:46 And then the information you give back
0:24:48 is not the answer to the user’s question.
0:24:52 It’s, hey, here’s a related question
0:24:55 that I think is very semantically similar to your question
0:24:57 with a doctor approved answer.
0:24:59 And then you use essentially your analytics, right,
0:25:02 to say, hey, how often and how far away
0:25:05 is the user’s query to the question bank that I have?
0:25:08 And then, you know, I can then go get more questions
0:25:10 that can have verified answers from doctors
0:25:13 and make that bank bigger and bigger and bigger over time.
0:25:16 And that’s how you actually get hallucination free
0:25:17 ’cause it’s a search, right?
0:25:22 So that golden set is the vetted questions and answers
0:25:24 that you’re then searching for to see.
0:25:25 – Well, that’s kind of–
0:25:27 – Users can’t see this, Sean made a face
0:25:28 and looked up, so I stopped.
0:25:29 – Oh, yeah.
0:25:30 So those are two different things.
0:25:31 – Okay.
0:25:33 – Like what I was talking about is what is the knowledge base?
0:25:35 It, you know, kind of a rag pattern.
0:25:36 – Yes.
0:25:38 – Golden answers is really the use case
0:25:39 specific evaluation frame.
0:25:40 – Okay, okay.
0:25:43 – And so you can think of it as most LLMs
0:25:46 look at like the MMLU as the benchmark
0:25:49 that they should be rated against,
0:25:51 which asks a bunch of multiple choice questions
0:25:53 for graduate students and things like organic chemistry.
0:25:56 That doesn’t tell you how to fix an AC.
0:25:59 It doesn’t tell you how to plant if there’s been a rainstorm
0:26:01 and you’re using this particular fertilizer
0:26:03 in the middle of, you know, Uganda.
0:26:07 For that, you need a different evaluation set, right?
0:26:09 And so that golden set is basically our answer
0:26:12 to how does somebody bring in their own use case
0:26:14 specific evaluation set?
0:26:16 And then we have a set of, you know, basically
0:26:18 you upload that was the question and answer pairs.
0:26:20 And then you say, here’s one version of the bot
0:26:21 using GPT-4.
0:26:23 Here’s one version using Gemini.
0:26:24 Here’s one version using Claude.
0:26:25 I’m gonna run them all.
0:26:28 And then what we do is we allow you to specify
0:26:30 and we have some default ones,
0:26:33 which answer is semantically most similar
0:26:34 to your golden answer.
0:26:35 And then we create a score out of that.
0:26:38 And then we just, you know, average that score
0:26:38 and then give you an answer.
0:26:39 That’s it.
0:26:42 And so this allows for a very flexible framework
0:26:44 for you to do your evaluation.
0:26:47 – Anything to, yeah, it was a long technical aside
0:26:47 of like, how do we–
0:26:49 – No, it’s, it’s good, it’s good.
0:26:53 – So we get the institute where looking at
0:26:56 how can we enable community specifically women
0:26:59 and minority genders to kind of define
0:27:03 what their own data set would look like.
0:27:05 How to create a data set that best represents
0:27:08 their community or their values.
0:27:11 How could they use those data sets to then create,
0:27:14 you know, fine-tuned models that enable others
0:27:18 within their community or outside to make imagery
0:27:20 and potentially animation even,
0:27:23 using those data sets that they have created.
0:27:26 And so that’s an exciting new project
0:27:28 that we’re gonna take off on this month.
0:27:31 And with Udav actually were looking at how,
0:27:35 and I think they kind of instigated the workspace feature
0:27:36 that we’ve kind of pulled out now,
0:27:39 which is how can we bring their young graduates
0:27:42 and even their PhD folks to start using AI tools,
0:27:46 quickly play with it without having to know
0:27:48 how to do the DevOps part.
0:27:51 I wouldn’t, it would take me another portion of my brain
0:27:52 to figure that out.
0:27:54 – I’m with you.
0:27:56 – So, you know, how do we make it possible
0:27:59 for like groups of people in their programs?
0:28:01 We’re looking at the DX arts program,
0:28:04 which is experimental arts program graduates
0:28:06 to be able to, you know, start creating stuff quickly
0:28:08 without all of the underlying stuff
0:28:12 that Sean eloquently and in great detail
0:28:13 has explained somehow.
0:28:17 – But also to do this in a collaborative way, right?
0:28:19 And I feel like that’s like the metaphor part
0:28:22 that will sort of get back into the AI workflow standards,
0:28:25 which is to say, you know, there was word around
0:28:28 for a long time and then we went to Google Docs
0:28:30 and we had a huge unlock of what it means
0:28:33 for to do real-time collaboration on document.
0:28:35 And you’re like, “Wow, I can be a lot more productive.”
0:28:37 – Sure, together. – Together.
0:28:41 – Look at like analytics and you take something like amplitude.
0:28:43 Amplitude say, “Well, you used to have data analytics
0:28:45 and like I ran a company where I would do
0:28:47 SQL training classes because I wanted
0:28:50 to democratize data analysis at night at my company.”
0:28:52 But then tabla or, you know, in the case of amplitude,
0:28:54 amplitude comes along and around.
0:28:56 And I can just share a URL with you,
0:28:58 which is like, you know, looking at our user analytics.
0:29:00 And if you want to change that from a weekly view
0:29:03 to a daily view, it’s just a dropdown, right?
0:29:05 And then, you know, Webflow arguably did the same thing
0:29:09 from like Photoshop, right, as a standalone desktop tool
0:29:11 to something that is collaborative in the client.
0:29:13 We think we can do the same thing
0:29:15 for the AI workflows themselves, right?
0:29:18 So that, again, we are working on these things
0:29:19 and I don’t have to worry about the underlying models
0:29:20 that are underneath them.
0:29:23 And you’re working at this higher level of abstraction
0:29:26 where I get to work and see outputs in a team environment.
0:29:28 And that’s very useful for learning,
0:29:29 which is the DX arts piece.
0:29:31 And, you know, it’s very useful
0:29:33 for improving frontline worker productivity.
0:29:35 And then as we make these things bigger and bigger,
0:29:37 you know, you want to do the same thing of,
0:29:39 hey, if I’ve got an image set
0:29:42 that we feel is underrepresented in something like Dolly,
0:29:44 I can take that image set and make my own model
0:29:46 and boom, suddenly make animation styles
0:29:48 around an indigenous art form, right?
0:29:49 That doesn’t exist there
0:29:50 ’cause the data doesn’t exist.
0:29:52 And that’s really the work that we’ll do with Gotay.
0:29:55 But it’s kind of like the same metaphors
0:29:57 keep getting built on top of each other.
0:29:59 And that’s the part that I think we find very exciting.
0:30:03 – Archnef, when you’re working with whether it’s women,
0:30:06 minorities, whatever sort of underrepresented community,
0:30:10 and, you know, particularly in a more rural place
0:30:13 where, again, you know, there’s access via phone
0:30:16 and things like trying to find a way to use Sora online,
0:30:19 right, just isn’t even in the, it’s a different perspective.
0:30:22 Are you finding that people are interested
0:30:26 and enthusiastic about not just learning how to use AI tools
0:30:29 but being represented in the data sets?
0:30:31 Is that something that you kind of have to explain
0:30:32 from the ground up?
0:30:34 And I’m asking in part because, you know,
0:30:36 talking about arts in particular, right,
0:30:38 and underrepresented communities, you know,
0:30:41 there’s been a lot of blowback in people talking about,
0:30:43 you know, being up underrepresented
0:30:47 or having their work used
0:30:49 without having been asked for consent.
0:30:51 And so kind of looking at the other side of it,
0:30:54 what’s the experience like in working with folks
0:30:57 who are coming from this totally different perspective?
0:30:58 – And thank you for that, Noah.
0:31:01 That’s a fantastic question, actually.
0:31:03 So I was, you know, recently in Manchester
0:31:05 and with friends at Islington Mill,
0:31:07 and we had a pretty deep conversation
0:31:09 pretty much around the same thing that you asked,
0:31:12 which is artists, creators definitely feel
0:31:13 there is a lot of pushback.
0:31:15 They have been exploited, their work,
0:31:17 their life’s works have been exploited.
0:31:19 Now, however, the cats out of the bag,
0:31:23 we’re not gonna be able to rewind some of this stuff,
0:31:27 but if we have to take kind of a peek into the future,
0:31:29 one of the missions I personally have
0:31:30 and feel very deeply about,
0:31:33 and I know that Goy is right there with me on that,
0:31:35 is that we’re kind of past the moment,
0:31:37 and like, you know, three years ago, four years ago
0:31:39 when we were doing the Radbots project,
0:31:41 it was, hey, can we enable the artists?
0:31:43 Can we give them the tools?
0:31:45 And then can they make what they would like to make?
0:31:47 I think we’re past that moment.
0:31:49 I think where we are at is
0:31:50 they need to make their own tools
0:31:53 and then make the things that they want to make
0:31:55 with the tools that are best servicing their needs.
0:31:59 That’s kind of where we’re at with Goy right now.
0:32:02 How do we enable people to make their own fine-tuned models
0:32:04 that allow them to, for example,
0:32:08 create imagery or animation that they would like to see,
0:32:10 that they would like to be represented with?
0:32:13 It’s just one example of how that could play out,
0:32:16 and I feel like there’s a significant urgency around that.
0:32:19 One is that in the making of those tools,
0:32:22 they get more aware, we all learn together,
0:32:25 and, you know, the workplace model is also very much that,
0:32:29 is that we learn better together, we make better together,
0:32:33 and the more we can get people, especially creative thinkers
0:32:36 and activists on this technology,
0:32:37 the better that world will be.
0:32:41 Absolutely, no, that’s great, absolutely.
0:32:44 So getting into kind of a last topic before we wrap up, standards.
0:32:45 Yes.
0:32:50 Sean, you were talking about the move from, you know, Word to Google Docs
0:32:52 and this collaborative environment.
0:32:56 HTML, obviously, is a great example of, you know,
0:32:58 a standard that has evolved, splintered,
0:33:01 would have you over time, but we all use the web, right?
0:33:05 How do you approach standards in this, you know,
0:33:06 new fast-moving world of AI?
0:33:09 So there’s always lessons from the past, right?
0:33:12 And so we hope so anyway.
0:33:13 We hope so, right?
0:33:15 We hope we learn the wisdom from the past.
0:33:17 But if you look at HTML,
0:33:20 HTML allowed for computer-to-computer communication
0:33:21 between networks, right?
0:33:23 But also had this other factor,
0:33:25 which I feel is completely under-appreciated,
0:33:27 which was view source, right?
0:33:29 Like the way that I learned to code
0:33:31 and figure out what HTML layout would happen
0:33:34 is ’cause I dissected the discovery home page.
0:33:36 And then, but there’s other ones that are kind of more recent
0:33:39 that I think are also indicative, like Kubernetes, right?
0:33:43 Google, like, you know, you rewind the clock 12 years,
0:33:46 Amazon had a lock on essentially cloud server configuration
0:33:49 and deployment, hence then Kubernetes came along
0:33:51 from an essentially upstart number two
0:33:53 and number three players like Google, right?
0:33:55 It was said, “Hey, I want to make it really easy
0:33:58 “to move from one platform to another.
0:34:00 “If I had a standard that could describe
0:34:02 “the configuration that I need,
0:34:03 “then suddenly you don’t have vendor law.”
0:34:06 And that has allowed the cloud infrastructure business
0:34:07 not be dominated by one company,
0:34:10 but to have, you know, there’s at least now big three
0:34:13 plus a bunch of local vendors globally.
0:34:15 And you can use the same Kubernetes file
0:34:17 to go and say, “This is what I need for all of them.”
0:34:19 So we think there’s a similar thing
0:34:21 around AI workflows and it already happens now.
0:34:23 Like you have tools like Open Router
0:34:26 that allows you to really easily switch your LLM,
0:34:29 but, you know, our take is if you can define
0:34:30 those kind of like high level interfaces,
0:34:32 like what’s an LLM do?
0:34:34 You put some text in, you get some text out.
0:34:36 Maybe you put some text and an image in
0:34:38 and then you get, you know, some text out,
0:34:40 maybe now some audio, right?
0:34:42 But, you know, you look at what is the interface
0:34:43 of a speech recognition model.
0:34:45 It’s like, well, you put some audio in
0:34:47 and maybe give it a language hint
0:34:48 and you expect some text out.
0:34:50 And then again, you want to swap, right,
0:34:51 for any model that’s underneath.
0:34:54 So part of it is there’s some standard interfaces
0:34:57 for these models and then those become steps.
0:35:01 And then you can compose those into essentially a chain,
0:35:03 a LLM chain or something like that,
0:35:05 but kind of a slightly higher level.
0:35:08 And then those steps end up becoming your recipe.
0:35:11 But the thing that travels with it is that golden data set.
0:35:12 So that allows you to say,
0:35:17 “Hey, I have my desired set of inputs and outputs
0:35:21 “and then I have my current set of steps that I should take.
0:35:24 “And then I can automatically just swap out the models
0:35:27 “as new ones are released and then boom, just tell you,
0:35:30 “you should really use this one, it’s better, cheaper, faster.”
0:35:31 And then that high level thing,
0:35:33 that is the AI workflow standard.
0:35:36 It’s basically like, what are your steps
0:35:38 extracted above the use of any given AI model?
0:35:40 Maybe you have a little bit of like,
0:35:41 what are the function calls that you’re going to expose
0:35:46 in there as well, kind of as, you know, open API configs.
0:35:47 Then what’s the evaluation set?
0:35:50 And our belief is if you had that higher level thing,
0:35:51 then you can take that and say,
0:35:53 “Oh, I want to run that on Claude
0:35:55 “or I want to run that on GPT builder.
0:35:58 “I want to run that on GUI or DeFi or Relevance.”
0:36:01 Then we suddenly have this, again, portable thing
0:36:02 that allows you to run.
0:36:05 – For folks listening and, you know, anybody,
0:36:07 but I want to kind of gear it towards that,
0:36:09 kind of new to the technology
0:36:13 or coming from less of a dev dev ops background
0:36:16 and more of a, you know, artist, activist,
0:36:19 writer type background or, you know,
0:36:21 the dev dev ops folks who are working with those people
0:36:25 who think it’s important to elevate those voices
0:36:27 and help them create the tools that they want to use, right?
0:36:30 What advice would you give to somebody out here listening
0:36:33 who thinks they have a new way to do it
0:36:35 or just wants to get involved with an organization
0:36:36 who’s doing it?
0:36:37 What would you tell them?
0:36:39 – Get started, get on GUI.ai.
0:36:40 It’s easy.
0:36:43 And if there’s any hiccups, contact us.
0:36:44 They’re very easy to catch.
0:36:45 And it’s not as hard.
0:36:48 It’s not as complicated as it feels.
0:36:49 There’s our platform.
0:36:50 There are others, too,
0:36:54 that are really trying to make these processes
0:36:57 simpler, faster, quicker, or more efficient.
0:36:59 And I don’t think there’s time to be wasted.
0:37:01 I think it’s now.
0:37:03 And there’s no point sitting in the sidelines
0:37:05 worrying about it or critiquing it.
0:37:08 Kind of got to get in there, make the stuff,
0:37:10 and then, possibly, make the barriers and the guardrails
0:37:12 that you need, as well.
0:37:15 You know, kind of take the bull by its horns.
0:37:15 – Yeah, excellent.
0:37:16 Yep.
0:37:20 The GUI.ai website, GUI.ai, is great.
0:37:23 Lots of use cases, lots of technical info, videos,
0:37:24 fantastic resource.
0:37:27 Are there other places you would direct listeners
0:37:28 to go in addition?
0:37:30 Social media, partner projects,
0:37:33 anywhere else besides the GUI.ai website.
0:37:34 And I’ll spell it while you’re thinking
0:37:37 G-O-O-E-Y for less fancy words.
0:37:38 Yeah.
0:37:41 I guess one thing that I’ll add to that is,
0:37:44 you can’t do good technology that changes the world
0:37:46 just by focusing on the technology, right?
0:37:49 That actually is just a means to the end.
0:37:53 And so, I think the thing for people to get started with
0:37:55 is, for me, it actually gets back to, like,
0:37:57 what’s the problem you’re solving?
0:37:58 Do you actually have something that looks like
0:38:00 golden questions?
0:38:01 And what does that mean?
0:38:03 It means, like, if you could imagine that,
0:38:06 hey, we could give great public defenders
0:38:10 for everyone in the country at no cost,
0:38:12 what would that look like, right?
0:38:13 What would be that set of expertise?
0:38:17 If we could say, hey, for any frontline worker,
0:38:19 I will be the nurse mentor for them,
0:38:21 helping them with triage and dealing with every
0:38:23 WHO guideline that they can imagine
0:38:25 and give them the right piece of advice
0:38:26 in their own language, right?
0:38:30 That is a real need for a real expert system.
0:38:32 And so, to think not so much of, like,
0:38:33 what’s the technology piece?
0:38:35 But what is actually the problem
0:38:38 where there’s a kind of expert out there right now
0:38:41 that’s expensive from a capacity-building perspective?
0:38:42 Right, right.
0:38:45 This is a place where AI can actually be really great,
0:38:47 which is we have collected wisdom from people
0:38:49 and processes and meta-processes,
0:38:51 all of 01 and documents and video.
0:38:53 And I feel like in the next year,
0:38:56 even with the current limitations we see around LLMs,
0:38:58 we can do this one well.
0:39:00 And so, for people, I would say you’d have to find
0:39:03 the problem worth solving in your community
0:39:04 or your business.
0:39:07 And say, if I could enable people to have that expert here,
0:39:10 they would earn more money, do their job better,
0:39:13 live longer, you know, have a better life.
0:39:15 And so, to focus not so much on the tech, but that part.
0:39:17 And then if you can get that, then, you know,
0:39:18 the tech tools are easy.
0:39:20 Arshna Prasad, Sean Blakesfett.
0:39:22 Thank you so much for joining the podcast,
0:39:24 telling us about GUI.AI.
0:39:27 I’ll say it again for the listeners, GUI.AI.
0:39:28 It’s easy. Check it out.
0:39:31 There’s so much to be done, so much you can do.
0:39:34 And thank you to folks like you who are making it easier
0:39:36 for more and more people to get involved,
0:39:39 be represented, and create the tools they need
0:39:40 to solve the problems they have.
0:39:41 Thank you.
0:39:42 Thank you.
0:39:45 [MUSIC PLAYING]
0:39:49 [MUSIC PLAYING]
0:39:52 [MUSIC PLAYING]
0:39:56 [MUSIC PLAYING]
0:39:59 [MUSIC PLAYING]
0:40:03 [MUSIC PLAYING]
0:40:06 [MUSIC PLAYING]
0:40:10 [MUSIC PLAYING]
0:40:13 [MUSIC PLAYING]
0:40:16 [MUSIC PLAYING]
0:40:20 [MUSIC PLAYING]
0:40:23 [MUSIC PLAYING]
0:40:26 [MUSIC PLAYING]
0:40:30 [MUSIC PLAYING]
0:40:33 [MUSIC PLAYING]
0:40:43 [BLANK_AUDIO]

0:00:10 [MUSIC]
0:00:13 Hello, and welcome to the NVIDIA AI podcast.
0:00:15 I’m your host, Noah Kravitz.
0:00:19 Digital humans and AI agents are poised to be big news this year.
0:00:21 They’re already making waves, in fact.
0:00:26 At CES 2025 in Las Vegas, Logitech G’s Stream Labs unveiled its intelligence
0:00:31 streaming assistant, powered by technologies from in-world AI and NVIDIA.
0:00:35 The intelligence streaming assistant is an AI agent designed to provide real-time
0:00:39 commentary during downtime and amplify excitement during high stakes moments
0:00:41 like boss fights or chases.
0:00:45 The collaboration brings together Stream Labs expertise in live streaming tools,
0:00:49 NVIDIA ace technology for digital humans, including AI vision models that can
0:00:53 understand what’s happening on screen, and in-world’s advanced generative AI
0:00:57 capabilities for perception, cognition, and adaptive output.
0:00:58 But what are digital humans?
0:01:00 Where are they going to be used?
0:01:03 And where are they going to make an impact in the enterprise and
0:01:05 in gaming and entertainment in particular?
0:01:08 And how does a designer design in the age of digital humans,
0:01:10 agentic AI, and beyond?
0:01:14 Chris Covert, director of product experiences in in-world AI,
0:01:18 who’s press release, by the way, or blog post, I paraphrased heartily from in
0:01:21 that introduction, so credit to.
0:01:24 Chris is here to dive into these points and more as we talk about some of the
0:01:28 technology that I think is really poised to really make a big impact this year
0:01:33 going forward and kind of shape this new area of digital experience for all of us
0:01:35 in what we’ll broadly call the AI age.
0:01:36 But at any rate, Chris is here.
0:01:41 So Chris, thank you so much for joining the AI podcast and welcome.
0:01:42 Thank you for having me today.
0:01:45 It’s a pleasure to be here, not only as a longtime partner of NVIDIA,
0:01:49 but also as a huge fan of your recent announcements in AI at CES this year as
0:01:53 well, genuinely amazing time to be talking about AI in this space.
0:01:54 It really is.
0:01:55 And so you were at the show.
0:01:56 You were in Vegas.
0:01:57 Couldn’t make it personally.
0:01:58 I’ll be doing dice.
0:02:00 I’ll be doing GDC, but our team was there.
0:02:02 We worked on the project.
0:02:03 All right, so let’s get into it.
0:02:08 Tell us a bit about in-world, in-world AI for listeners who may not know.
0:02:11 We can get deeper into the assistant in particular,
0:02:14 a little later in the conversation, if that’s cool, just so you can kind of set
0:02:18 the stage for all of us about what it is that we’re going to be talking about.
0:02:20 Yeah, so I, you know, extremely biased.
0:02:22 I have the best job in the world.
0:02:25 And in-world, we make the leading AI engine for video games.
0:02:28 And I get to work with the most creative minds in the industry,
0:02:31 most creative minds in the world of gaming and entertainment to answer the
0:02:35 question, how can we make fun more accessible to the people that make our
0:02:36 favorite experiences?
0:02:40 And I’m not going to tell you it sounds like a great job.
0:02:42 It is, it is.
0:02:43 And I’m not just blowing smoke here.
0:02:47 And we’ll definitely, as we go through this, you know, emphasize fun,
0:02:50 accessible in people, and why I think those are the most important here.
0:02:54 But in-world’s mission is to be that AI engine for this industry, right?
0:02:58 The more that the technology progresses, the lower the barrier
0:03:00 to entry to make AI experiences.
0:03:04 But we find there are still challenges, regardless of if you’re a studio
0:03:08 with a massive staff or you’re an AI kind of native company building
0:03:12 up these experiences from grassroots, that there’s a massive gap
0:03:14 between prototyping and deployment of AI systems.
0:03:19 And then to do that and find the fun for a user for an experience
0:03:20 is still incredibly challenging.
0:03:24 So we offer not only the platform, but the services to help you
0:03:26 create that fun in a deployable manner.
0:03:31 So kind of briefly run down maybe what you guys talked about at CES this year
0:03:34 and then we’ll put a pin in the assistant, like I said, and come back.
0:03:35 Yeah, that’s awesome.
0:03:38 So blog posts is out there, fantastic videos out there.
0:03:41 A lot of demonstrations of what happened, but succinctly, you know,
0:03:45 to plug the collaborative effort here in World NVIDIA and Streamlabs
0:03:50 got together for CES and we put to the test this, you know, streaming companion
0:03:52 that could not only act as a cohost to a streamer,
0:03:54 but also serve to support the creator in other ways,
0:03:58 like handling production for them on their stream as well.
0:04:01 We showed a demo of this happening at the conference, always a risk
0:04:04 because you’re playing in a big space with low, you know, internet bandwidth.
0:04:08 But we played this over Fortnite and the streamer, the demo where in this case
0:04:11 and their agentic cohost, you know, chatting during the game,
0:04:15 grilling each other for bad rounds, reacting to trends that are happening in chat,
0:04:17 cheering together when you manage to get kills.
0:04:21 But then interestingly enough, you know, when chat or when the streamer wants a replay,
0:04:24 streamer has to just ask the agent, they’ll clip it,
0:04:27 they’ll set up the screen for that replay and they’ll keep that stream feeling fluid.
0:04:30 So it’s one of those use cases, again, trying to align on like,
0:04:35 where do agentic systems actually play in enterprise and this entertainment industry
0:04:37 is we’re finding they’re hyper specialized.
0:04:40 And in this use case, we have this perfect, you know,
0:04:43 one streamer has to wear all of these different hats during their stream.
0:04:45 How can an agentic system come assist them
0:04:48 so that they can focus on making the best content
0:04:51 and all of the other things in managing that content can be done for them.
0:04:54 So back in the day, and this is part of why we got to come back to it.
0:04:58 I did a lot of YouTube stuff and we tried to do some live streaming.
0:05:00 This was in like the early mid 2000s.
0:05:05 And I remember setting up, you know, open source plugins to try to get graphics
0:05:06 over the live stream.
0:05:10 And then I had like a secondary camera, you know,
0:05:13 like a phone with a camera with a USB cable and all this stuff.
0:05:14 But that sounds amazing.
0:05:17 So I’m excited to talk about that to kind of then back into it
0:05:19 or start with the technology and go forward.
0:05:23 A genetic AI cohost digital humans.
0:05:24 Let’s talk about those things.
0:05:27 You want to start with digital humans because we’ve talked about a genetic AI
0:05:30 and we’ll get to it and it has the different, you know, what an agent is or isn’t.
0:05:33 I think it’s still a little bit malleable and context specific.
0:05:37 But when we talk about digital humans, when InWorld talks about digital humans,
0:05:38 what are we talking about?
0:05:39 You know, that’s such a great question.
0:05:41 And it brings up an important distinction.
0:05:44 When most people say digital humans, they’re thinking of chatbots.
0:05:48 Text-based tools that respond with mostly pre-programmed scripts
0:05:52 to help a specific task or guide users through a set of questions or instructions.
0:05:56 But at InWorld, we focus on AI agents that go far beyond simple chat.
0:05:59 These agents can autonomously plan, take actions,
0:06:01 and proactively engage with their environment.
0:06:07 Unlike a typical chatbot, which waits for a user to have some kind of question
0:06:12 or response in and then that AI agent provides a canned response,
0:06:15 our agents at InWorld are designed to interpret multiple types of inputs,
0:06:19 whether that’s speech or text or even vision or sensor data.
0:06:22 And then dynamically think and respond in real time.
0:06:23 They don’t just answer questions.
0:06:25 They can initiate tasks.
0:06:26 They can adapt to different contexts.
0:06:28 And they can carry out sophisticated interactions
0:06:31 that make sense within the context of what you’re doing.
0:06:34 So if you look at where digital humans are used mostly today,
0:06:35 and again, I’m super biased,
0:06:38 it’s often in the high volume chatbot or, you know,
0:06:40 personal digital assistance space.
0:06:43 But if you consider where they can have the biggest impact,
0:06:45 that’s when they function as these truly autonomous agents.
0:06:48 Planning and acting and proactively helping people
0:06:50 in ways that traditional chatbots can’t.
0:06:52 It’s not question and answer.
0:06:54 It’s answer or question that I didn’t even know I had yet.
0:06:56 And that’s the core of our focus at InWorld,
0:06:58 building AI agents that can not only look human,
0:07:02 but can think and react and solve problems like one,
0:07:05 using multiple modalities to interact with their world.
0:07:06 In our case, a digital one.
0:07:09 When it comes to building with agentic AI,
0:07:13 do you feel even like you have kind of a solid vision
0:07:15 of what is to come even near term?
0:07:20 Or is it still very early days of figuring out everything from,
0:07:23 you know, trying different types of infrastructure
0:07:26 to, you know, sort of the design and higher level thinking
0:07:28 about how to architect these things?
0:07:28 That’s a great question.
0:07:32 And the reality is we’re shifting toward an agentic AI framework,
0:07:33 as I mentioned before.
0:07:37 And shifting toward that framework is no longer really an option.
0:07:40 It’s becoming essential for us and our partners.
0:07:44 When you need any sense of autonomy or kind of real time
0:07:46 or runtime adaptability, latency optimization,
0:07:48 any sense of custom decision making,
0:07:52 you need a robust framework that lets you take control of that stack.
0:07:55 No one platform is going to serve the needs of our industry
0:07:56 as standalone.
0:07:58 So building a framework that is flexible,
0:08:00 that is robust to the needs and growing needs,
0:08:03 changing needs, super diverse needs of our industry
0:08:05 is incredibly important.
0:08:06 What we’re seeing more and more of is enterprises
0:08:09 that want to own that architecture end to end.
0:08:11 What I mean by that is they need the flexibility
0:08:14 to decide what models to use, where to use them,
0:08:16 how data is fed into their systems,
0:08:18 how to manage compliance and risk,
0:08:21 all of these factors that require really custom logic
0:08:22 and custom implementations.
0:08:24 So it’s adopting this, you know,
0:08:28 agentic AI framework that really puts the power in their hands.
0:08:30 It’s not just about future proofing.
0:08:33 It’s about giving them full control over how AI evolves
0:08:35 within their organization at any given time, right?
0:08:37 The industry is going to move quickly.
0:08:38 We want to make sure our partners can move
0:08:41 just as quickly as they get inspired.
0:08:45 But to me, you know, my favorite key differentiator here,
0:08:47 Future of Inworld, why the framework model,
0:08:50 why all of these changes is that such a framework
0:08:52 doesn’t just guarantee technical flexibility.
0:08:54 That’s great, and that’s going to make a lot of people happy,
0:08:57 but it also opens up the creative design space as well, right?
0:09:00 By collaborating with people that are thinking
0:09:02 of the most outrageously beautiful,
0:09:05 outrageously wonderfully weird experiences
0:09:07 you could possibly imagine and giving them tools
0:09:09 that are opened up to really let them craft
0:09:11 their own AI architectures.
0:09:14 We’re really, you know, helping push the boundaries
0:09:18 of what and how we can deliver innovative experiences, right?
0:09:21 We made the jokes earlier, but if we say that conversational AI,
0:09:22 you know, this like chat bot nature,
0:09:26 is now feeling very entry level, it’s feeling baseline,
0:09:28 then what we’re doing with this agentic framework
0:09:30 is building out that blueprint for the future
0:09:32 so that our partners and our customers
0:09:37 can help imagine, build, and deploy AI experiences
0:09:40 that genuinely didn’t feel possible even six months ago, right?
0:09:41 Taking that to a whole new level.
0:09:44 If we go more than like two years out,
0:09:47 I’m at a 10% confidence level of where this is actually going to be.
0:09:50 Yeah, I appreciate the candor.
0:09:52 Yeah, it’s just, you know,
0:09:54 and I’ve been doing this a couple of times in my career throughout
0:09:57 where I’ll work with somebody in an advisory capacity
0:10:00 where we’re looking at the future three to five years out
0:10:02 and by the time we finish our analysis,
0:10:04 the thing that we had two years out is now open sourced
0:10:06 and you’re like, cool, so back to the drawing board
0:10:09 because it’s just the industry moves so quickly
0:10:13 that what is possible, it’s actually really hard to nail down
0:10:16 and we will get into kind of, hopefully in this conversation,
0:10:20 how designing around AI and these systems is that challenge
0:10:23 because I think we’ve learned some fantastic lessons
0:10:25 that hopefully people listening can take away as well.
0:10:29 I really see there, you know, when it comes to trying not to sound too obvious
0:10:31 when we’re talking about agents,
0:10:34 the evolutionary chain of how an agent will grow,
0:10:36 how an agent, you know, the technology behind it
0:10:38 will proceed in even the near future.
0:10:41 It comes all down to agency and again, it’s agents.
0:10:44 It sounds obvious, but what I really mean by that
0:10:46 is the complexity of that cognition engine
0:10:49 and I’ll try and crack this net with an analogy
0:10:51 and I’ll try and make it fast because I could talk about this.
0:10:52 Now, take your time.
0:10:53 We’ll take your time with the analogy anyway.
0:10:58 We have this, you know, first again is that conversational AI phase
0:11:00 and I’ll use a gaming analogy, right?
0:11:04 The conversational AI phase gives avatars, gives agents,
0:11:05 I’ll use them interchangeably today,
0:11:08 extremely little agency in doing anything other than speaking, right?
0:11:12 It may be able to respond to my input if I ask it to do something,
0:11:14 but it’s not going to physically change the state of something
0:11:17 other than the dialogue it’s going to tell me back.
0:11:19 So, I think in terms of an analogy,
0:11:21 like a city-building simulation game,
0:11:24 in this phase, conversational AI,
0:11:25 I can ask it about architecture
0:11:27 and it could give me some facts about architecture.
0:11:30 I can ask it about plans I have to build this little area
0:11:31 and it could give me some advice.
0:11:34 I could ask it, “Hey, how do I limit the congestion
0:11:37 or how do I maximize traffic over this little commercial area?”
0:11:41 And it could tell me in words how to potentially do those things,
0:11:43 but it won’t be able to do things like place buildings
0:11:46 or reconstruct roads or even select the optimal facade
0:11:48 for the style of the region
0:11:50 because that’s completely out of its scope.
0:11:50 And not to…
0:11:51 It can’t act.
0:11:52 Yeah, it can’t act.
0:11:53 It can’t act at all.
0:11:53 It is…
0:11:55 Okay, just making sure that I’m fine.
0:11:55 Yeah.
0:11:58 It is a very rudimentary perception,
0:11:59 kind of what is it seeing,
0:12:02 what context does it have and cognition,
0:12:04 how is it planning, how is it reasoning,
0:12:06 very simple engines that drive that.
0:12:08 Really, I’m going to talk to it.
0:12:10 It’s going to say, “Oh, you know,
0:12:12 this person cares about architecture.
0:12:14 I’m going to pull everything I know about architecture,
0:12:16 turn that into a nice response,
0:12:19 weave together kind of their question into my response,
0:12:22 and then boom, now we have simple conversation AI.”
0:12:24 But not too far ahead of that.
0:12:26 And I say that lovingly and endearingly
0:12:29 is a simple task completion AI.
0:12:31 It’s very much still call and response.
0:12:33 You have an action engine.
0:12:34 You say, “I’m having an agent
0:12:37 that can only construct new buildings
0:12:38 exactly where and when I tell them.
0:12:40 I can say place a building at this intersection
0:12:41 and it does it, boom.
0:12:43 I can say change the building to this type
0:12:44 and boom, it does it.”
0:12:44 Right.
0:12:46 And while you argue that
0:12:47 that requires a little bit more perception,
0:12:49 maybe a little bit more reasoning,
0:12:50 definitely some action,
0:12:53 it’s limited to an extremely small set of actions
0:12:56 and it’s pretty much just scripting with extra steps.
0:12:58 It knows what it’s going to expect.
0:12:59 It’s waiting for me to say it in a way
0:13:02 that it can map to that is the place of building action.
0:13:04 I’m going to place a building done.
0:13:08 If a language model provides the description,
0:13:11 including like I wish I could change the background
0:13:13 to mountains or whatever, right?
0:13:15 Or envision a background with mountains
0:13:16 and it can’t act, right?
0:13:17 Can the system be automated?
0:13:18 What you’re talking about
0:13:23 so that the language model can tell the action cognition model?
0:13:24 What to do?
0:13:25 I forget what term you used.
0:13:28 I have a large action model in my head,
0:13:30 which yeah, I have thoughts about that terminology,
0:13:32 but I don’t know if they’re my own thoughts or not.
0:13:34 So we’ll get into it.
0:13:36 But you’re definitely hinting at, and again,
0:13:38 I should have upfront said this.
0:13:39 There are four phases.
0:13:41 We are at phase two.
0:13:43 I think a lot of virtual assistants are at that phase two.
0:13:46 What you’re describing in kind of intent recognition
0:13:49 or kind of passive intent recognition even
0:13:51 is there are no formal names for this,
0:13:54 but I’ll call this like an adaptive partner phase
0:13:56 where the AI is observing
0:13:58 and responding to changes on its own, right?
0:13:59 And this is the natural evolution
0:14:02 of where a lot of AI systems are heading today.
0:14:03 CES, a good example of that.
0:14:05 I have no doubt throughout the year,
0:14:07 you’ll get really close to getting here
0:14:09 as another core standard.
0:14:10 I think right now,
0:14:12 a lot of the kind of simple task completion AI
0:14:15 is a core standard across enterprise and gaming.
0:14:16 This adaptive partner phase
0:14:20 where this agent in this analogy, extended analogy,
0:14:22 would notice changes like newly built roads
0:14:25 or new residents influx into this tiny area
0:14:28 and automatically adapt construction plans.
0:14:30 It’s not micromanaging every decision,
0:14:33 but it feels like you’re collaborating with an agent
0:14:36 or a unit that has just enough context
0:14:37 to make smart decisions on its own,
0:14:40 like an evolution of a recommendation engine
0:14:43 being driven by a cognition engine here.
0:14:44 So it’s not just learning,
0:14:47 but it feels like it’s learning what we need
0:14:48 even before we ask it.
0:14:51 And I expect to see, again, a lot of this in the next year.
0:14:53 Again, I think that’s phase three.
0:14:55 I think there’s still a phase four.
0:14:57 And I think that’s a fully autonomous agent.
0:15:00 And that stage, again, continuing our analogy,
0:15:02 is a player two, right?
0:15:04 Where player three is it’s adapting to us.
0:15:07 Stage four is, hey, this thing is an agent all on its own.
0:15:10 It feels like I’m playing against another human.
0:15:11 It is making decisions that feel optimal
0:15:14 to its own objectives, alike to mine or not.
0:15:18 And I think this is where a lot of people think
0:15:19 about agentic AI.
0:15:21 They think we are there today.
0:15:24 There’s quite a big gap in the deployment of these systems
0:15:26 to get to that truly autonomous agent.
0:15:29 That is kind of the gold rush right now
0:15:32 is creating a cognition and a perception
0:15:34 and an action engine ecosystem
0:15:38 that can feel natural to a user.
0:15:39 I’m speaking with Chris Covert.
0:15:43 Chris is Director of Product Experiences at N-World AI.
0:15:45 I made him kind of, I made us kind of put a pin
0:15:47 in their product announcement to kind of bring this
0:15:50 into the, from the abstract a little bit into the concrete.
0:15:52 I’m glad you humor me, Chris, because I think kind of digging
0:15:54 into, well, what does this stuff actually mean?
0:15:55 And it’s sort of moving.
0:15:57 And one of the things I should say it’s kind of moving
0:15:58 and evolving.
0:16:00 Maybe the lingo is moving, targeting,
0:16:03 but the tech and what’s actually happening is evolving
0:16:04 and still so fast.
0:16:04 All right.
0:16:07 So let’s talk about the big announcement from CES.
0:16:10 You said you wanted to talk through kind of the design
0:16:12 process a little bit and how you make these choices.
0:16:14 So if that’s something that makes sense to you to do
0:16:17 in the context of talking about the streaming assistant,
0:16:18 great.
0:16:20 And if not, take it whichever way you want to go first.
0:16:21 That sounds perfect.
0:16:21 Yeah.
0:16:23 So I’ll be honest around here, right?
0:16:26 In any demo that you see publicly,
0:16:28 especially when you have more than one company demoing it,
0:16:31 there’s a lot of design, but also a lot of urgency
0:16:33 to show something quickly and cool.
0:16:37 So of the things that I’ll say are design axioms
0:16:40 that I love to live by, we had to sacrifice some
0:16:42 just to make sure that something awesome could
0:16:44 be demonstrated at CES.
0:16:46 Putting a timeline on creative vision is always challenging.
0:16:47 But–
0:16:47 Right.
0:16:49 And that’s like you can’t– we live in–
0:16:51 I mean, I don’t build software for a living.
0:16:52 You do.
0:16:55 But we live in such an age of like, we’re past this state,
0:16:58 where beta became, all right, everybody ships beta.
0:16:59 And then we got used to it.
0:17:01 And it’s like, right, there’ll be updates
0:17:03 and downloadable content and this and that.
0:17:05 But not CES.
0:17:05 That’s the date.
0:17:06 Show up or don’t.
0:17:07 So–
0:17:08 Exactly.
0:17:08 Exactly.
0:17:11 But in terms of design, and this might sound odd,
0:17:13 coming from a tech startup on the call here today.
0:17:16 But my first piece of advice, regardless of who we’re
0:17:18 working with, is always to ignore the technology
0:17:20 and focus solely on the experience
0:17:22 that you really want to create.
0:17:24 Where is the value in the experience?
0:17:25 The why?
0:17:28 And as soon as you can lead, or not as soon as you can lead
0:17:31 with the tech, but as soon as you do lead with the tech,
0:17:32 hey, I have an idea.
0:17:34 What if we had an LLM that could?
0:17:36 You already are risking thinking too narrowly
0:17:38 and missing bigger opportunities.
0:17:41 And I genuinely would love to say the reason behind this
0:17:42 is because it’s human-centered.
0:17:45 I come from the intersection of autonomous systems, AI,
0:17:46 and human-centered design.
0:17:49 So I would love to say that it’s a human-centered approach
0:17:50 to design, and that’s why I do it.
0:17:53 But in reality, is that the technology just advances
0:17:55 so quickly by the time your idea goes into production,
0:17:58 that tech-grounded idea that you had now
0:18:00 looks like a stone wheel compared to the race car
0:18:01 you thought it was six months ago.
0:18:05 So again, like the urgency behind some of these things
0:18:08 is challenging, especially when you come from a tech-grounded
0:18:10 design principle standard.
0:18:14 So I’m a firm believer in the moonshot first approach,
0:18:17 where you begin by clarifying why you want to build something
0:18:20 before how you decide to build it, especially for anything
0:18:22 with a shelf life greater than just a few months at this stage.
0:18:24 Again, if you start with how you end up
0:18:26 with a bunch of low-hanging fruit,
0:18:28 and a bunch of low-hanging fruit makes for a really bad smoothie.
0:18:32 So what is a moonshot experience?
0:18:35 Like, I often refer in design thinking workshops
0:18:38 to the classic impact versus feasibility matrix.
0:18:41 And this isn’t like a very exciting conversation.
0:18:42 So I’ll try and keep it high energy.
0:18:44 But typically, when you’re looking at impacts–
0:18:46 I was just saying the smoothie.
0:18:48 Sorry, I couldn’t let the smoothie go unacknowledged,
0:18:50 but I didn’t want to interrupt it.
0:18:51 Well, we’ll come back to smoothie right for now.
0:18:52 OK, good.
0:18:55 Because again, this is that it’s in that quadrant,
0:18:58 or it’s in that matrix, impact feasibility.
0:19:00 Your low-hanging fruit ideas are the things
0:19:03 that are going to be low feasibility or low impact.
0:19:06 But typically, you’re aiming for the upper right quadrant,
0:19:09 which is the highest feasibility and the highest value,
0:19:11 where impact and feasibility are so high,
0:19:13 the ideas feel easy to build
0:19:15 and have a lot of inherent value in them.
0:19:17 But what I’ve experienced when working with partners
0:19:20 and their AI ambitions is that that’s actually a trap,
0:19:23 that that quadrant right there is doomed for failure.
0:19:28 Because often the ideas that are the ones worthy of pursuit
0:19:31 the most are the ideas that almost feel impossible,
0:19:34 but if real would deliver extraordinary impact.
0:19:35 Right, that’s right.
0:19:37 As you get closer to building that value,
0:19:40 and you’re going from idea to prototype to implementation,
0:19:45 the technology is likely grown in not only capability,
0:19:47 efficiency, but also accessibility in ways
0:19:51 that probably outpace our own kind of traditional lens
0:19:54 of what is feasible and our imagination of where the tech can go.
0:19:56 So that sounds like a lot of nothing.
0:19:58 So my actual advice is,
0:20:01 when we’re designing these types of experiences,
0:20:03 start from that moonshot that feels impossible
0:20:06 and then break it down into a functional roadmap.
0:20:09 What you saw at CES in this demo with Streamlabs
0:20:11 and NVIDIA and in-world technology
0:20:13 creating the streamer assistant
0:20:15 is actually just Horizon 1, that insight.
0:20:17 What can we build that gives us insight
0:20:20 into whether this is something that provides value?
0:20:22 Where does it provide value and to who?
0:20:25 The thing we demonstrated at CES
0:20:30 was just the first pass proof of concept of this capability.
0:20:34 The kind of upper potential of where this can go,
0:20:35 where we want this to go,
0:20:37 and where we’re going to explore its value
0:20:42 is still very much architected out toward a much longer roadmap.
0:20:45 And again, showing what we showed at CES
0:20:47 and what I love about this industry right now
0:20:50 is we’re able to show really compelling integrations
0:20:53 with very meaningful use cases in this industry.
0:20:55 I’m not about to wax poetic,
0:20:58 like our industry is the same as healthcare
0:20:59 or things like that and enterprise,
0:21:02 but showing digital companions, virtual assistants,
0:21:05 digital humans at the state that we’re showing them now,
0:21:08 knowing what’s to come is an incredible place
0:21:10 to look ahead and say, “Okay, cool.
0:21:13 This scratches an itch that either the market needed
0:21:16 or that technologically is doing something
0:21:17 that could never be done traditionally before.
0:21:23 Where this moves in the next six months to six years
0:21:24 is anyone’s game.”
0:21:27 Okay. So two questions, but they’re related.
0:21:29 One’s easy. The other’s a follow-up.
0:21:32 What is the availability of the assistant?
0:21:34 It was demo state very early.
0:21:36 Do you have kind of a roadmap for that?
0:21:38 I shouldn’t say availability of a roadmap.
0:21:40 And that might be the answer to the second part,
0:21:44 but what else is on the horizon for you for in-world AI?
0:21:46 I mean, you talked a little bit before
0:21:49 about the broader horizon for agentic AI
0:21:51 and avatars and assistants and such,
0:21:52 but you can go back there if you like.
0:21:53 What’s coming up this year,
0:21:56 kind of maybe near-term that you’re excited about?
0:21:58 Well, man, this year being near-term
0:22:01 is funny at InWorld because we move very quickly.
0:22:03 This year is many near-terms stacked against each other.
0:22:06 Well, right. And I’m figuring we’re taping now,
0:22:08 but it’s going to go live.
0:22:10 You know, there’s a little bit of a buffer
0:22:12 and we reference CES.
0:22:14 Yeah. So we certainly have productization ambitions
0:22:15 for this demo.
0:22:18 What we’re doing post-CES is we’re taking the feedback
0:22:21 of what we’ve built and we’re augmenting it in many new ways.
0:22:24 Again, what was built for the demo was a proof of concept.
0:22:27 As many proof of concepts on show floors are,
0:22:29 it wasn’t robust to all the inputs we would want.
0:22:30 It wasn’t robust to all the games
0:22:32 that we would want it to be playable with.
0:22:34 So we’re trying to build out that ecosystem
0:22:35 in an intelligent strategic way
0:22:37 so that if it were to go to market,
0:22:40 it would be usable to as many streamers as possible
0:22:42 who wanted to leverage this type of technology.
0:22:46 So keep your ears on the beat for what’s about to come out.
0:22:50 I have no doubt that between Nvidia, InWorld, and Streamlabs,
0:22:52 all announcements and all possible show floors
0:22:54 that we can show our advancements on
0:22:56 will be shown at the right time.
0:22:57 So super exciting for that.
0:22:59 As it relates to InWorld, oh boy,
0:23:01 it’s such a fascinating question
0:23:03 because a lot of what we’re doing,
0:23:04 we’re doing with partners
0:23:07 with such fascinating game development lead times
0:23:09 that I hope that we’ll be playing
0:23:11 more in-world driven experiences
0:23:13 in the next, you know, when this releases,
0:23:15 but also over the next four or five years,
0:23:17 depending on the scale of the company we’re working with.
0:23:21 So I’m genuinely excited for what’s to hold
0:23:22 as our platform develops.
0:23:25 Again, as we’ve seen in just the last year or so
0:23:27 with more and more competitive players
0:23:30 in the space of providing AI tools,
0:23:32 and then just fantastic partners
0:23:35 that are helping this industry become more accessible
0:23:37 to people of all different backgrounds,
0:23:41 really hope to see the, I wouldn’t say consolidation of tools,
0:23:43 but the accessibility, I’ll keep using that word,
0:23:47 the accessibility of different AI platforms.
0:23:50 Hey, I want to use this model, but in this engine,
0:23:51 to become a lot easier.
0:23:53 And InWorld’s goal is certainly to make that happen
0:23:55 for as many industries as we can,
0:23:58 in particular the gaming and entertainment industry,
0:24:00 but it doesn’t happen without partners like Nvidia
0:24:02 and the work that you guys are doing with ACE.
0:24:05 So where I, you know, where I think InWorld is going
0:24:08 is to make that easier, to continue to work with studios
0:24:11 to find the fun and to convince every player.
0:24:14 And in particular, you know, let me be honest,
0:24:16 the YouTube commenters that, hey,
0:24:19 there is actually a world here where this technology
0:24:21 is not only fun and immersive,
0:24:23 but it’s something that the entire industry views
0:24:25 as it’s gold standard.
0:24:27 So I think we’re there a lot sooner than we think.
0:24:29 I think it’s right around the corner,
0:24:33 but could not be more excited to continue
0:24:35 to work with creatives to help them tell stories,
0:24:39 to help them, you know, flex and use their imagination
0:24:42 as much as possible to make the best possible experiences.
0:24:43 We do it every day.
0:24:45 We may not see it in the near term
0:24:47 because games take a while to make,
0:24:49 but genuinely excited.
0:24:51 Excellent. Well, I’ll offer more fun.
0:24:53 The world always needs more fun to counterbalance everything else.
0:24:56 Chris, for listeners who would like to learn more
0:24:59 about anything specific we might have talked about
0:25:01 or just broadly about InWorld,
0:25:02 where can they go online?
0:25:05 Website, obviously, but are there social handles?
0:25:08 Is there a separate blog or even research blog?
0:25:09 Where would you direct folks?
0:25:10 Yes, to all of the above.
0:25:13 You can find all of that at inworld.ai.
0:25:17 And we have our blog for partners and experiences there.
0:25:18 We have technical releases.
0:25:19 We have research done.
0:25:21 All of our socials are linked there.
0:25:22 If you want to stay up to date
0:25:24 with all the announcements that we make,
0:25:26 we have a lot of fun and we like to talk about it.
0:25:28 So definitely stay up to date.
0:25:30 Fantastic. Well, thanks for taking a minute
0:25:31 to come talk about it with us.
0:25:32 We appreciate it.
0:25:35 And best of luck to you and InWorld AI
0:25:36 with everything you’re doing this year.
0:25:39 And maybe we can do it again down the road.
0:25:41 I love it. Thank you so much, Noah. Appreciate it.
0:25:44 [MUSIC PLAYING]
0:25:48 [MUSIC PLAYING]
0:25:51 [MUSIC PLAYING]
0:25:55 [MUSIC PLAYING]
0:25:58 [MUSIC PLAYING]
0:26:02 [MUSIC PLAYING]
0:26:06 [MUSIC PLAYING]
0:26:09 [MUSIC PLAYING]
0:26:13 [MUSIC PLAYING]
0:26:16 [MUSIC PLAYING]
0:26:20 [MUSIC PLAYING]
0:26:23 [MUSIC PLAYING]
0:26:26 [MUSIC PLAYING]
0:26:30 [MUSIC PLAYING]
0:26:40 [BLANK_AUDIO]