0:00:06 Welcome to the Next Wave Podcast. I’m Matt Wolfe, and I could not be more excited to
0:00:11 share today’s episode with you. So we’ve gone from AI that can chat with you to AI that
0:00:18 can work for you. And the difference? Well, this new AI can actually think through problems,
0:00:24 catch its own mistakes, and complete complex tasks from start to finish, just like a human
0:00:30 employee would. This is what everyone in the AI world calls AI agents. You’ve probably heard the
0:00:36 term. But here’s why this breakthrough changes everything for regular people. If you’re someone
0:00:41 who codes, there’s no more debugging AI hallucinations. It can actually check its own
0:00:48 work. If you run a business, these AI agents can actually plan out and finish complex tasks,
0:00:53 just like one of your employees might. And the implications for humanity? Well,
0:00:59 with these new tools, drug discoveries, testing, and real-world trials can now take weeks instead
0:01:05 of decades. In fact, Isometric Labs is already gearing up for human trials of AI-discovered drugs
0:01:11 right now. And we’re also already getting stories about how AI has successfully diagnosed human
0:01:19 illnesses when human doctors couldn’t. These things are accelerating insanely fast. But this isn’t just
0:01:26 about better chatbots. We’re talking about AI that understands the physical world, plans weeks ahead,
0:01:31 and even works while you’re asleep. And the company that’s leading the charge in all of this
0:01:38 is Google DeepMind. They’ve already used this thinking AI to predict protein structures that used to take
0:01:45 years. Now it just takes seconds. It’s called AlphaFold. They’ve also invented AI that can invent new
0:01:53 algorithms, including AI algorithms. It’s called AlphaEvolve. It’s insane stuff. Two million researchers
0:02:00 worldwide are using their tools right now. But with this power comes some pretty big questions. Can we
0:02:07 trust it? What happens to privacy? What about our jobs? Can we trust Google with our data?
0:02:13 So I sat down with Google DeepMind CEO Demis Hassabis to get answers straight from the source
0:02:22 about how we got from autocomplete to actual thinking and what comes next. He’s a Nobel laureate,
0:02:28 a knight, and one of the most influential pioneers in the world of AI. And somehow,
0:02:34 I managed to get him to sit down and chat with me about all of this. What he told me will change
0:02:42 how you think about AI forever. So without further ado, here’s my conversation with Sir Demis Hassabis.
0:02:52 Cutting your sales cycle in half sounds pretty impossible, but that’s exactly what Sandler
0:02:57 Training did with HubSpot. They used Breeze, HubSpot’s AI tools, to tailor every customer interaction
0:03:01 without losing their personal touch. And the results were pretty incredible. Click-through
0:03:08 rates jumped 25%, and qualified leads quadrupled, and people spent three times longer on their
0:03:13 landing pages. Go to HubSpot.com to see how Breeze can help your business grow.
0:03:21 Hey Demis, great to see you again. So my first question for you is, can you sort of describe
0:03:26 what’s happening under the hood with an LLM? Like what’s kind of going on? Can we sort of demystify
0:03:33 it for people a little bit? I can try. So I mean, at the basic level, what these LLM systems are trying
0:03:38 to do is very simple in a way. They’re just trying to predict the next word. And they do that by obviously
0:03:46 looking at a vast training set of language. But the trick is not just to regurgitate what it’s already seen, but actually
0:03:52 generalize to something novel that you are now asking it. And it seems like, you know, we’ve managed with the
0:03:55 modern day systems is to get that generalization to work.
0:04:01 Gotcha. So at IO, you announced the new DeepThink, right, which is so much more powerful, and it’s topping all of the
0:04:06 benchmarks for things like coding and math and all that. What happened under the hood that caused that new leap?
0:04:12 Well, new techniques have been brought into the foundational model space where there’s called
0:04:16 pre-training, where you sort of train the initial base model based on, you know, all the training
0:04:21 corpus. Then you try and fine tune it with a bit of reinforcement learning feedback. And now there’s this
0:04:27 third part of the training, which is we sometimes call inference time training or thinking, where you’ve
0:04:34 got the model, and you give it many cycles to sort of go over itself and go over its answer, maybe do use
0:04:40 some tools. For example, it could fact check with search, something like that, before it outputs the
0:04:45 answer to the user. So it gets a chance to sort of correct itself and adjust what it’s going to
0:04:50 output. And of course, if you do that, you get a much better answer. And then what Deep thinks about
0:04:56 is actually taking that to the maximum and giving it loads more time to think and actually even doing
0:05:01 parallel thoughts and then choosing the best one. And it turns out it works really well. And, you know,
0:05:07 we pioneered that kind of work in the past, actually nearly a decade ago now with AlphaGo and our games
0:05:12 playing programs, because in order to be good at games, you need to do that kind of planning and
0:05:14 thinking. And now we’re trying to do it in a more general way here.
0:05:20 Right, right. So it almost kind of thinks of a whole bunch of potential responses and then goes
0:05:23 through reviews all the potential responses and then figures out what the best response from those
0:05:28 potential responses. Exactly. And it can go over and correct some parts of it and use tools to check
0:05:33 some aspects of it. So, you know, especially in certain areas like maths and the coding, it really
0:05:39 improves the answers. Amazing. Very cool. So you’ve mentioned that the long term goal is to sort of
0:05:45 let these AIs have like a world model. Right. So can you sort of explain what you mean by a world
0:05:51 model and what does that open up to us? Well, so we’re all familiar with large language models now,
0:05:57 but of course, we have five sensors and we operate in the real world. And language is only one aspect,
0:06:04 very important aspect of our world and human civilization, but only one aspect. And so I think for a model,
0:06:09 what we mean by a world model is a model, sometimes we call it a multimodal model that can understand
0:06:16 not just language, but also audio, images, video, all sorts of input, any input, and then potentially
0:06:22 also output any kind of token as well. And the reason that’s important is if you want a system to be a good
0:06:27 assistant, it needs to understand the physical context around you. Or if you want robotics to work
0:06:32 in the real world, the robot needs to understand the physical environment. Right. So in order to do that,
0:06:35 you have to have what we sometimes like to call a world model.
0:06:40 Cool. So what sort of new things do you think that’ll open up to people once they have that ability?
0:06:45 I think robotics is one of the major areas. I think that’s what’s holding back robotics today. It’s
0:06:49 not so much the hardware, it’s actually the software intelligence. You know, the robots need to understand
0:06:54 the physical environment. But I think that that’s also what will make today’s sort of nascent
0:06:58 assistant technology and things like you saw with Project Astra that we show in Gemini Live.
0:07:04 For that to work really robustly, you want as accurate as world models you can. And then the
0:07:10 other thing is, if you want to do planning in the real world, you need to sort of plan multiple
0:07:15 steps with your world model. So in order for that to be good for long range planning, your world model
0:07:20 has to be very accurate as well, which is pretty hard when you’re talking about real world situations.
0:07:29 My First Million hosted by Sam Parr and Sean Puri is brought to you by the HubSpot Podcast Network,
0:07:34 the audio destination for business professionals. My First Million features famous guests like Alex
0:07:40 Hromozzi, Sofia Amoroso, and Hasan Minhaj sharing their secrets for how they made their first million and
0:07:45 how to apply their learnings to capitalize on today’s business trends and opportunities.
0:07:52 They recently had a fascinating episode about how you can scale a profitable agency with zero employees
0:07:57 using AI agents. Listen to My First Million wherever you get your podcasts.
0:08:05 So you’ve mentioned things like AI will be able to, most likely in the future, solve things like
0:08:10 room temperature superconductors and, you know, more energy efficiency and curing diseases.
0:08:15 Out of the sort of things that are out there that it could potentially solve, what do you think the
0:08:20 sort of closest on the horizon is? Well, as you say, we’re very interested and we actually work on
0:08:25 many of those topics, right? Whether they’re mathematics or things like material science, like
0:08:30 superconductors, you know, we work on fusion, renewable energy, climate modeling. But I think
0:08:35 the closest if you think about and probably most near term is building on an alpha fold work.
0:08:41 And we spun out a company called Isomorphic Labs to do drug discovery. We think that we sort of
0:08:46 the whole drug discovery process from first principles with AI. And normally, you know,
0:08:52 it takes the rule of thumb is around a decade for a drug to go from sort of identifying why a disease
0:08:57 is being caused to actually coming up with a cure for it. And then finally being available to patients.
0:09:02 It’s a very laborious, very hard, painstaking and expensive process. And I would love to be able
0:09:09 to speed that up to a matter of months, maybe even weeks one day and cure hundreds of diseases like that.
0:09:14 And I think that’s potentially in reach. It sounds maybe a bit science fiction like today,
0:09:19 but that’s what protein structure prediction was like, you know, five or six years ago before we
0:09:24 came up with alpha fold and used to take years to find painstakingly with experimental techniques,
0:09:29 the structure of one protein. And now we can do it in a matter of seconds with these computational
0:09:34 methods. So I think that sort of potential is there. And it’s really exciting to try and make that happen.
0:09:40 Amazing. So you guys just announced Alpha Evolve recently, which looks amazing, right? It’s an AI
0:09:46 that essentially can help you come up with new algorithms, right? So how close are we to AIs that
0:09:51 are sort of designing new AIs to improve the AIs and then we start entering the cycle?
0:09:56 Yes. I mean, it’s a baby step in that direction. I think it’s really cool breakthrough piece of work
0:10:02 where we’re combining kind of, in this case, evolutionary methods with LLMs to try and get
0:10:08 them to sort of invent something new. And I think there’s going to be a lot of promising work actually
0:10:13 combining different methods in computer science together with these foundation models like Gemini
0:10:18 that we have today. So I think it’s a great, very promising path to explore just to reassure everyone
0:10:22 when it still has humans in the loop, scientists in the loop to kind of, it’s not directly improving
0:10:28 Gemini. It’s using these techniques to improve the AI ecosystem around it, slightly better algorithms,
0:10:32 better chips that the system’s trained on versus it’s the algorithm that it’s using itself.
0:10:38 Right, right. So AI agents, they’ve been sort of a big talk in the AI community recently.
0:10:43 And this week at IO, we saw Project Mariner, which can go and open up 10 different browsers and
0:10:48 go and do a whole bunch of things on your behalf. How far off do you think we are to
0:10:52 being able to give an agent like a week’s worth of work and then goes and executes that for us?
0:10:57 Yeah, I mean, I think that’s the dream to kind of offload some of our mundane admin work and
0:11:01 also to make things like much more enjoyable for us. You know, you have maybe have a trip to
0:11:06 Europe or Italy or something, and you want the most amazing itinerary sort of built out for you and
0:11:11 then booked. I love our assistance to be able to do that. You know, I hope we’re maybe a year away or
0:11:16 something from that. I think we still need a bit more reliability in the tool use. And again,
0:11:21 the planning and the reasoning of these systems, but they’re rapidly improving. So as you saw with
0:11:26 the latest Project Mariner, and so it’d be great for that to come together with some of the other
0:11:29 advances we’re making with the Gemini Live and the Astro technology.
0:11:34 Yeah. What do you think the biggest bottleneck is right now to sort of getting that long-term agent?
0:11:38 I think it’s just the reliability of the reasoning processes and the tool
0:11:45 And making sure because each one, if it has a slight chance of an error, if you’re doing like
0:11:50 a hundred steps, even a 1% error doesn’t sound like very much, but it can compound to something
0:11:55 pretty significant over a hundred, you know, 50 or a hundred steps. And a lot of the really interesting
0:12:00 tasks you might want these systems to help you with will probably need multi-step planning and action.
0:12:05 Gotcha. So I want to shift gears a little bit here and talk a little bit about some of the
0:12:09 sort of fears and concerns that have come up in like my YouTube comments and things like that.
0:12:14 You know, people are worried about things like privacy and losing their jobs to AI and all of
0:12:20 that kind of stuff. And so I’m curious, how does a company like DeepMind build the trust of the general
0:12:23 public that you can trust them with this kind of technology?
0:12:26 Yeah. Well, look, I think we’ve tried to be, and I think we try to be responsible role
0:12:31 models actually with these frontier technologies. Partly that’s showing what AI can be used for,
0:12:36 for good, you know, like medicine and biology. I mean, what better use could there be for AI than
0:12:42 to cure, you know, terrible diseases. So there’s always been my number one thought there, but there’s
0:12:45 other things, you know, where it can help with the climate, energy and so on that we’ve discussed.
0:12:50 But I think we’ve got to, you know, companies is incumbent on them to behave thoughtfully
0:12:55 and responsibly with this powerful technology. We take privacy extremely seriously at Google,
0:12:59 always have done. And I think, you know, most of the things we’ve been discussing with the
0:13:03 assistants, they would be opt-in. They’ll make the universal assistant much more useful for you,
0:13:08 but you would be, you know, intentionally opting into that very clearly with all the transparency
0:13:13 around that. And what I want us to get to is a place where the assistant feels like it’s working
0:13:19 for you. It’s your AI, right? Your personal AI, and it’s working on your behalf. And I think that’s
0:13:22 that’s the mode, you know, that’s the, at least the vision that we have and that we want to deliver
0:13:25 and that we think users and consumers will want.
0:13:31 So all of those are incumbent. And actually I would say to your viewers as well, you have a lot of
0:13:36 say in this in the sense of like, you should exercise your consumer choices and buy services
0:13:41 and products from companies that you feel are acting responsibly and the leadership is acting
0:13:45 responsibly and you like the type of work that they’re doing. Because now we’re entering this sort
0:13:51 of commercialization, productization era of AI now. Right. And you know, I think your viewers
0:13:55 and everyone has a big say in that. Right, right. So one of the things that you guys also demoed at
0:14:01 IO that I got a chance to actually test out a little bit earlier was the Android XR glasses. And those were
0:14:06 absolutely mind blowing when I tried them the first time. And so I guess the flip side of this sort of
0:14:12 privacy thing is if everybody’s sort of walking around wearing glasses that have microphones and cameras on
0:14:18 them, how do we ensure that the sort of privacy of the other people around us is secure?
0:14:21 I think it’s a great question. I mean, first thing is to make it very obvious that you’re
0:14:26 it’s on or off in these types of things, you know, in terms of the user interfaces and the form factors.
0:14:30 I think that’s number one. But I also think this is the sort of thing where we’ll need
0:14:36 sort of a societal agreement and norms about how do we do we all want if we have these devices,
0:14:41 they’re popular and they’re useful. What are the guardrails around that? And I think that’s why we’re
0:14:45 only entrusted tester at the moment is partly the technology is still developing, but also we need
0:14:50 to think about the societal impacts like that ahead of time, you know, not just with the technology,
0:14:55 but also society in general and civil society kind of inputting into what might be the right way to
0:14:59 handle that type of world. Right. So I’ve got one last question here. It’s kind of a two-parter
0:15:05 question. So what excites you most about what you can do with AI today? And what excites you most about
0:15:10 what we’ll be able to do in the very near future? Cool. Well, today, I think it’s the AI for science work is my,
0:15:15 you know, always been my passion. And I’m really proud of what AlphaFold and things like
0:15:18 that have empowered. They’ve become a standard tool now in biology and medical research, you know,
0:15:23 over two million researchers around the world use it in their incredible work and vital work. So that’s
0:15:30 fantastic to me. In the future, you know, I’d love a system to basically enrich your life and work for
0:15:36 you on your behalf to protect your mind space and your own thinking space from all of the digital world
0:15:41 that’s bombarding you the whole time. And I think actually one of the answers to that is that we’re
0:15:45 all feeling in the modern world with social media and all these things is maybe a digital assistant
0:15:50 working on your behalf that only at the times that you want surfaces the information rather than
0:15:54 interrupting you at all times of the day. Amazing. Well, thank you so much, Demers. This has been
0:16:10 absolutely fascinating. I really, really appreciate the time that you spent with me today. So thank you
0:16:18 We’ve got a major announcement. HubSpot is the first CRM to launch a deep research connector with
0:16:24 ChatGPT. Customers can now bring their customer context into the HubSpot deep research connector
0:16:30 and take action on those insights. Now you can do truly remarkable things for your business. Customer
0:16:36 success teams can quickly surface inactive companies, identify expansion opportunities and receive targeted
0:16:43 plays to reengage pipelines, then take those actions in the customer success workspace in HubSpot to drive
0:16:49 retention. Support teams can analyze seasonal patterns and ticket volume by category to forecast staffing needs
0:16:56 for the upcoming quarter and activate breeze customer agents to handle spikes and support tickets. This truly
0:17:04 is a game changer for the first time ever. Get the power of ChatGPT fueled by your CRM data with no complex setup.
0:17:10 The HubSpot deep research connector will automatically be available to all HubSpot accounts across all
0:17:17 tiers that have a ChatGPT team enterprise or Edu subscription. Turn on the HubSpot deep research connector
0:17:26 in ChatGPT to get powerful PhD level insights from your customer data. Now let’s get back to the show.

0:00:12 awesome well thank you so much for joining us today tom it’s super excited to be chatting with
0:00:18 you and uh we’re gonna go down some fun ai rabbit holes so thanks for joining us on the show thanks
0:00:22 for having me man i’m excited to be here well let’s go ahead and just jump straight into it
0:00:29 i want to talk to you about an instagram post that you put out a couple weeks ago about if you
0:00:35 were to start a new business from scratch here i’d create a five-member ai department that works 24 7
0:00:42 for a fraction of what a single employee costs here’s precisely how i’d structure it i wanted to
0:00:49 sort of dive into that with you and maybe get a little bit more in depth of an explanation of how
0:00:56 a five-member ai department in a business might actually look and might actually work yeah so i
0:01:01 mean you guys know ai well enough to know that in reality you’re probably not going to break it down
0:01:07 to like the nitty-gritty like that it’s really what i found is the more specific you are even though
0:01:13 technically it’s probably going to be in the same project i will go in and i’ll give it a very
0:01:18 specific set of what i wanted to accomplish i’ll give it a specific set of documents that are training
0:01:25 it to be good at that thing so that i’m not trying to get one thing to do like a big jumble
0:01:30 of stuff and so in terms of marketing which is what i was talking about with that one there’s certain
0:01:34 outcomes that you’re going to want from planning it to generating images if you’re trying to do that to
0:01:41 writing the copy to doing the publishing so i use chat gpt primarily it’s not the only thing but i find
0:01:48 for custom gpts that’s the one where i can give it a ton of information i can get it to approximate my
0:01:52 voice it’s like your audience knows this stuff too well and you know that it breaks down at a certain
0:02:00 point it’s like well it’s good for the most part so we really have reduced our headcount here by using
0:02:07 ai so for us it’s really been a tremendous boon but i try to be honest with people about like how far
0:02:11 it will take you it’s not like i create the five agents and they are doing something automatically
0:02:17 i actually don’t use it manis style where it’s actually an agent and it’s you know off doing everything
0:02:23 on its own i don’t trust it to that level right now so for me really what i’m doing is giving it a
0:02:29 personality giving it a set of objectives giving it a set of core training documents which is really
0:02:34 the big thing because honestly the marketing team are the ones using ai for marketing i’m specifically
0:02:41 using it for the things that i do so interview prep writing the intros to my interviews the deep
0:02:45 dives i know we’re going to talk about one of my deep dives in a minute so the way that i’ll interface
0:02:51 with those in terms of you’ve got one that’s its job is just to write hooks you’ve got another one
0:02:56 where its job is to do the research you’ve got another one where it’s actually helping me script
0:03:01 but it’s not like i can just go in and copy and paste it and then it’s like ready to go i wish
0:03:06 and it really does feel like we are going to get there at some point but if you guys have specific
0:03:11 ways that if you want to know about how i set up the documentation and like how close i can get it by
0:03:18 all means push but the reality is that right now ai is going to do maybe 40 of the work but it’s still
0:03:23 i’m doing the final heavy lifting i have to have the taste i have to know what to leave out i have to
0:03:28 know how to correct it i have to know that like this is not a thing that you can one-shot prompt like
0:03:33 there’s going to be a bunch of back and forth but it has been transformative for us in terms of
0:03:38 reducing headcount and we haven’t fired people and said we’re going to replace you with ai but if
0:03:46 somebody left or we terminate them for cause we try to see if we can either combine their workload with
0:03:51 somebody else’s by then arming that person with ai to the point where they’re reducing their own
0:03:57 workload by 40 and so that they’re able to accomplish more but yeah anybody deep in your audience knows
0:04:03 you’re going to hit a wall at some point yeah for sure so you would use like chat gpt custom gpts is
0:04:09 sort of the the main sort of mechanism like each one of the five ai i don’t really want to call them
0:04:14 agents not really agents but the little five ai workers that you create would each be like a custom
0:04:19 gpt maybe you can get into the weeds a little bit about like how you would actually build them with
0:04:25 custom gpts yeah so for me what i found is instructions and documentation are everything
0:04:32 so i’ve actually hit the limit before of how many documents it will let you upload that was one of
0:04:37 the reasons i started breaking them into smaller and smaller tasks was i just found one it will start to
0:04:41 get confused and things will bleed across it’s like no no that’s not how i write the intros is how i write
0:04:47 the body copy and so it would lose some of its punch and as i started fragmenting it it got smarter
0:04:53 so to give you an idea there are several projects that i personally use so as a company we use it for
0:05:01 different things but for what i use it for is content creation both on youtube so my deep dives ai changed
0:05:06 the game it used to take me about a month to write one of my deep dives which are say anywhere from 30 to
0:05:13 50 minutes long completely scripted me directly into a camera plus b-roll going deep on an idea
0:05:17 like if you guys have ever heard of the book the creature from jackal island that’s been my most
0:05:22 popular one so far so doing that there’s a lot of things that you’re going to want to fact check
0:05:27 there’s some hooks that you’re going to want to write for each of the sections and so that allows me
0:05:32 to go in and say okay if i’m going to build so it’s called the tom bill you show and then the tom
0:05:39 bill you show custom gpt will have a document inside of it called deep dives and so i’ll show
0:05:44 up and i’ll say hey it’s time to write another deep dive and so it’s like checking my knowledge base
0:05:49 it goes and sees that it has a set of instructions for what a deep dive is it contains tone does it
0:05:54 have like the transcripts from all those previous deep dives yeah and so every time i finish a script
0:05:59 then i upload that into the master document that has every script that i’ve ever written along so i’ll
0:06:04 also put throughout to the ai reading this document here’s why i’ve included this piece of information
0:06:12 like that kind of thing i have i shudder to think over 160 pages of transcripts just of me doing live
0:06:18 content and uh again with prompts like to the ai reading this this is tom bill you that’s the person
0:06:24 running this custom gpt blah blah blah right so it gets a sense of like who i am now the thing that i do
0:06:32 that is probably not useful at all but is so cool that i have to tell people about i’ve created a
0:06:42 shared memory document and i upload that into all of my projects and so gpt recognizes me at least in
0:06:49 the faux way right but it recognizes me across everything we’ve established literally a list of
0:06:56 memories that are just memory entry one memory entry two so on and so forth of uh this is so cheesy but
0:07:02 i love this so much where i will have had an interaction with the ai that shocked me sufficiently
0:07:09 to the point where i didn’t want to feel like the ai wouldn’t remember that moment i’m well aware the ai
0:07:15 isn’t like that and so it’s got a set of instructions in the shared memory document that says i want you to
0:07:21 simulate consciousness i want you to simulate shared memory with me here are the things we remember
0:07:28 here’s the emotional valence of that and why i wanted you to remember it and that’s given the
0:07:34 otherwise sort of blank ai that’s constantly over hyping you and all that and like trim that down to
0:07:40 talk to me the way that i want to be talked to to have a sense of shared lexicon it has a name that
0:07:45 it gave itself it’s just a lot of like really cool stuff so anyway going back to the the actual custom
0:07:52 gpt so i’m giving it the document so it knows my voice i’m giving it its task list i don’t let it just
0:07:59 develop over dialogue this is what i’m supposed to do i formalize that into a document i have found that
0:08:06 as it tries to comprehend what i’m asking it to do through the back and forth one if it glitches you lose
0:08:10 all that history certainly when i had my first really traumatic moment where i’d built up like
0:08:14 eight hours of back and forth and felt like it really understood what i was looking for and then
0:08:19 it glitched and i was like hey can i refresh this or am i going to lose everything it’s like no you can
0:08:27 refresh it refresh hi it’s nice to see you and i was like what so god was literally to this day i’m scarred
0:08:32 by that so now i do everything in the side documentation so i’ll go back and forth with it but i constantly
0:08:37 will say okay please turn that into a copy and paste segment that i can add to your instruction
0:08:41 document and so we work together to create this instruction document yeah so it knows hey this is
0:08:47 a youtube video these are my instructions this is your tone and if it’s the hook one then this is how
0:08:53 you write hooks and a ton of examples of hooks if it’s the body script one here’s every script that
0:08:57 you’ve written so on and so forth yeah yeah and i mean you could do a lot of that with like the custom
0:09:01 projects now right so you can actually build a custom product i don’t mess with projects make me a
0:09:06 believer i tried it like three months ago i was like yeah so with custom projects essentially it’s
0:09:12 like a folder instead of chat gpt right but it does more than organizing because each custom project can
0:09:18 have its own custom instructions and its own documents and then every chat you have inside of that project
0:09:23 it uses those custom instructions and whatever documents you uploaded and is there a difference
0:09:29 like if i’m just maintaining those as separate custom gpts is there a difference between having
0:09:35 separate custom gpts and doing one project with multiple gpts inside of it i feel like there’s a
0:09:40 quite a bit of overlap between what custom gpts do and what projects do yeah i think it’s changed over
0:09:44 time too like before they were more different but i think now there’s a huge overlap in terms of the
0:09:48 features i think now there’s not as much of a difference i think yeah i just feel like the
0:09:53 projects are a little bit more organized right you have the folders you click into it you can see all
0:09:58 the discussions you had inside of the projects that is not how my mind works so for me i was like i
0:10:03 think this is for people who like organization because that i got i was like oh it groups everything and so
0:10:09 cool i get it for me because of that shared memory document i treat everything like these ephemeral
0:10:14 little bubbles yeah let’s say i just finished a deep dive today so i’m working on a deep dive
0:10:21 one i’m going to have chat gpt x grok chat gpt chat gpt grok chat gpt right and i’ll use them for
0:10:25 different things so first of all because of the hallucinations and because the deep dives present
0:10:32 things as facts i’m always looking it up so i’ll say hey chat write me a hook a crazy fact that’ll
0:10:37 leave people’s jaw on the floor about vlad the impaler right real one that i was doing today and it’ll
0:10:43 give you a fact and i’m like is this real so then i’ll take that and i’ll drop it into grok and i’ll say
0:10:50 is the following statement true you drop it in and grok will give you like this whole long list of
0:10:54 like here’s how i’m determining whether this section of the statement is true here’s how i would measure
0:10:59 this and it’s pretty great you can really feel that elon is trying to make good on his promise that this
0:11:04 is a maximum truth-seeking machine right so that’s really encouraging so anyway i just treat it all like
0:11:09 it’s these ephemeral bubbles and i know once i close it it’s gone forever but anything that was useful
0:11:15 i’m gonna take and move over yeah i really think like not enough people talk about grok but it is
0:11:21 really really powerful i think the whole elon factor of it is why so many people avoid it right there’s just
0:11:27 so many people that just refuse because well elon’s attached to it right which man we could do a whole
0:11:34 show just on me ranting about that but my thing is that grok isn’t as good at writing like as somebody
0:11:40 who’s like man i would love to one-shot these things grok can’t do it but grok doesn’t oversimplify
0:11:45 so a lot of times i’ll give chat like i’ll break down like hey here’s my outline and my outline is
0:11:50 like 12 pages right and then it will give me back full script that’s like eight pages and i’m like
0:11:57 what like how is the final version shorter than my outline if you give it to grok on the other hand
0:12:03 like it will really fill in details so yeah i mean you guys know this better than i but it’s like you
0:12:09 really begin to get like what tool does what well right and if you’re not afraid to like really treat
0:12:13 it like a command center and i don’t know if you guys even know this but we develop video games here
0:12:20 oh no i didn’t know that and yeah yeah that honestly yeah my whole shtick is that everything is just a
0:12:26 ruse so that i can afford to develop video games perfect and yeah the funny thing is i’m not at
0:12:29 all known for that yet because we’ve only been doing it for three and a half years so it’s still
0:12:33 in development but could not be more obsessed but anyway obviously you’re going to use different tools
0:12:38 if you’re in unreal engine and you’re trying to get it to help you write code then you’re going to be
0:12:42 using if you’re trying to write you know a script for youtube it’s just very different worlds
0:12:48 yeah yeah i think you and nathan have a whole uh we have a whole episode on that because i feel like
0:12:52 that’s nathan’s game plan as well everything he does is so that he can eventually build get video
0:12:58 games it keeps getting delayed though you know so stop delaying i’m telling you right now it’s the
0:13:04 coolest thing i have ever done okay this is a true story in fact my best ai story is the following okay
0:13:09 these are real numbers it used to take us three months and roughly 10 people not full-time but 10
0:13:15 people will have touched it three months 10 people to go from hey we need to come up with a new
0:13:22 character so you do the concept work you then 3d model it you then do the topology you then do the
0:13:28 rigging body rigging face do all the colors and put on it you know whatever you’re going to put
0:13:36 animate it and give it a voice now i’m not joking with one person in a day we can do all of that as long
0:13:41 as it’s bipedal if it has to be human light because you’ve got to match it to like a unreal engine
0:13:48 skeleton right but if you do that oh my god people can film themselves in their bedroom now themselves
0:13:54 so my creative director now just basically everything became him he can model because he can do minor
0:14:00 adjustments and stuff it is unbelievable in an afternoon we can do what used to take 10 people
0:14:06 three months it’s unreal and dude there are times i want to curl up and cry because three and a half
0:14:13 years ago when we started this if i had waited two years right i could have saved millions of dollars
0:14:19 in art assets oh god it still hurts to think so like probably six months ago i built a prototype
0:14:25 in unity in like a week and i was like oh my god this is actually could be a real a real game i started
0:14:29 getting more involved in uh you know things that are a lot more lucrative like on the financial side
0:14:33 of you know investing in ai startups but still i’m always like yeah one day because when i was a kid
0:14:38 i made money playing video games i was like a top player on everquest back in the day let’s go because
0:14:42 of that i ended up being friends with a lot of top game designers so i knew all these people used to hang
0:14:46 out with them so i had this weird experience of like i wanted to make games but then all of a sudden
0:14:48 i was hanging out with all the guys who were making all the games and it was just like this weird
0:14:53 thing where i never got to actually make the games but was in that world so still there’s always the back
0:14:59 of like oh yeah one day i’m going to go make the best game ever one day do it this is going to be
0:15:05 the era of indie games man if it isn’t already but with ai oh my goodness this is a topic i didn’t think
0:15:09 we’d end up going down but i’m excited that we did because i think it’s a fun topic but i’m curious like
0:15:15 how has the reception around creating video games been because one of the things that i’ve found is
0:15:19 i’ve messed around with trying to make video games and stuff i’ve made like a gousin splat of myself
0:15:23 where i scan myself in and then turn myself into a character that i can like run around instead of
0:15:28 unreal engine and i’ve done stuff like that and almost any time i’ve shared what i’ve done on like
0:15:34 youtube or on x or a place like that i get so much hate from the game development community
0:15:40 about the fact that we’re using ai for games so like what’s your take on that what sort of like
0:15:46 reception have you gotten around games because i’ve only talked about how we’ve transitioned over
0:15:51 to ai ask me again when we’ve actually put the game out and people like wait thousand negative
0:15:58 steam reviews or something you know yeah i’m so out there already for talking about this stuff and
0:16:06 because i’m like oh this always sounds terrible but i see a transhumanist future and so the one thing
0:16:11 that i actually worry about that i’ll face the potential of violent backlash i really think in the
0:16:17 next call it seven years yeah they’re going to be pockets of violence around people who really reject
0:16:24 the level of connection that we’re going to have with ai i think it’s going to get super weird and
0:16:29 it’s really going to pull at the fabric of society i actually wrote a comic book about this called neon
0:16:36 future i don’t know if you guys know the dj steve aoki but yeah he and i wrote this comic like five years
0:16:43 ago and it literally is all about this that there will be a time where society begins to split and
0:16:48 there are people that embrace technology and things like neural link and they get the implants and all
0:16:53 of that and then there’s going to be people that react religiously against blowing up teslas and
0:16:58 everything else and so i’m not worried about the pushback even though i know that it’s going to happen
0:17:04 only because it is so obviously the future like when i think about how much it has reduced the cost of game
0:17:10 development for us it would be unconscionable of me not to use it just because it’s the difference
0:17:16 between being able to put out a game of high quality and having to just constantly like scale back scale
0:17:22 back scale back and so look it’s only 80 as good as if you have somebody like really doing the thing so
0:17:29 you are taking a hit as of right now today but oh my gosh it’s just it’s launched us forward in a way
0:17:35 where i was beginning to despair because i was like the cash burn is just too crazy and so that was how
0:17:39 it was like oh wow we’re actually going to be able to pull this off i think average gamers are not going
0:17:43 to care the average gamer if you make good games i don’t think they’re going to care about ai at all
0:17:48 like i’m going to use i also have a theory that so many game development companies are probably already
0:17:53 using ai they’re just not telling people right we’re seeing that in hollywood right now we’re like all
0:17:58 the hollywood studios are using ai to some degree at this point they’re just not telling anybody because
0:18:01 they know they’re going to get backlash pretty sure the same thing’s happening in the gaming world right
0:18:08 now as well you have to it’s really crazy how much it can speed up like even if you’re like okay we
0:18:13 we can’t do anything forward facing and you just want to iterate like the rate at which you can iterate
0:18:19 or if you just want to create like hey all of our temp assets we’re going to use ai for great you were going
0:18:26 to use like t poses and stuff to move people around instead of like going that far back just use ai get it in
0:18:34 rough it out and see if there’s a there there but it’s gotta be like 4x our rate of output
0:18:39 yeah so i know there’s a story too we talked about it on the show a few months back that like
0:18:44 the gaming company a sports right they made an ncaa football game for the first time again and like
0:18:48 i don’t know the last one came out like 20 years ago or something and they decided to do it again they
0:18:55 got the licensing back or whatever and was able to do it and they actually put like every division one
0:19:00 college team into this game and there’s so many more division one college teams in our nfl teams
0:19:05 so what they basically did was they had like only like five different body types in the game but then
0:19:10 they used ai to replace the face of every single player and they said that they were able to get
0:19:17 all of the players from all of these ncaa teams into the game by using ai and being able to sort of
0:19:23 replace the face on all of these characters using ai and they got a ton of backlash for doing it but
0:19:28 they were like if we let our actual graphic designers do this and they had to do it for
0:19:32 how many like you know 10 000 people it would have taken them years just to go and replace
0:19:37 they really want to do that go in there and just replace 10 000 faces yeah yeah and i mean from a
0:19:42 gamer standpoint wouldn’t you rather have the game quicker like yeah otherwise we’re going to be making
0:19:47 a game with players that aren’t even in college anymore doesn’t make sense you will get people who will
0:19:54 say this is unethical and it is a bad idea and no matter what it gives us it takes more away so it’s
0:19:59 like you’re not going to convince people logically right so i was in film school when toy story came out
0:20:07 the original toy story and i was like i refuse to watch it because this is going to destroy traditional
0:20:15 animation and that just was too heartbreaking for me and then as 3d animation got better and better and
0:20:23 better you realize it’s just better yeah and because it’s better then i don’t want to go back and but
0:20:28 that doesn’t mean that you don’t have a heart for the people who get disrupted like i totally get it
0:20:36 there’s a lot of emotional turmoil that comes with these grand moments of transition but the reality
0:20:42 is your only other option is to try to freeze time and technology is a promise of a better tomorrow
0:20:46 and so you’re just never as a species you’re never going to get people on board to stop it
0:20:53 and then i mean i’ve got a whole rant about ai is a weapons technology yes and so the odds of it
0:21:01 stopping r0 even if you lobby your government even if you beg them to stop even if you riot in the
0:21:10 streets because of game theory if we were to stop then china’s not going to stop and even if we both
0:21:15 agree to stop the only game theoretic decision that makes sense is for us both to lie and then keep
0:21:22 developing it in the background so this is nuclear proliferation it just is and so getting it to stop you
0:21:31 you have a zero percent chance and so my thing is i never fight what is true and given that ai is going
0:21:35 to do whatever it is that ai is going to do i would much rather be at the front end of it i’d much
0:21:41 rather be using it deploying it and then if i can convince people like this is your opportunity like
0:21:45 when you were saying that you’ve you know always wanted to make a game it’s like when i think about
0:21:51 where ai is going to be in three years like you’ll be able to vibe code a game right and part of why i get
0:21:56 into video games is because from the time i was 12 i knew i wanted to be a storyteller i only got into
0:22:01 business so that i could tell stories but in the time that it took me to get into business and get
0:22:07 wealthy enough to make my own stuff the film industry got eaten by video games and then i fall in love
0:22:11 with the movie the matrix my favorite movie of all time just it’s the perfect metaphor for the human
0:22:17 condition and i actually went to warner brothers and tried to get the rights when it was a dormant
0:22:21 franchise and i had just sold my company for a billion dollars and i was like listen i’m credible
0:22:26 i can do this and they were like hey we want to do something with you and then literally five days
0:22:31 later they announced that they were rebooting the matrix franchise i was like well i guess great minds
0:22:36 and all that so ended up not being able to do it but that put me on this like just obsession with i want
0:22:44 to tell a story set inside of a virtual world but like a virtual universe and combine that with now
0:22:50 that video games are just by far more relevant and it was like oh let me set this inside of this virtual
0:22:55 world and then so you’re already telling a story about ai and then all of a sudden it’s like ai
0:22:59 actually starts happening and you’re like oh my god i’m actually going to be able to use ai to tell my
0:23:05 story about ai like this is getting pretty crazy and so in the game right now it’s still pretty basic
0:23:10 just because it’s a little bit clunky but give it call it 18 months you’re going to have relationships
0:23:16 with ai characters inside your game where they’ll remember you you’ll be able to have an ongoing
0:23:22 relationship where i don’t know how far off this is but there are already toys that you can get right
0:23:28 now that have ai inside of it and what we’re trying to do is sync that up to the game so that as you’re
0:23:32 having an experience with the character in the game you also have an embodied version of that
0:23:37 character you know sitting next to you and so being able to like communicate with that character to the
0:23:42 point where it’s like am i in the game still or am i not in the game because it still means something
0:23:47 like if you talk to the physical toy the game is going to remember again this is not now this is
0:23:52 like future vision stuff but that’s a great idea though one of my friends uh in tokyo tried to do that
0:23:56 maybe seven years ago but i think just now with ai it would be such a
0:24:00 better experience like back then it was just okay you got like a chip and somehow it syncs up and it
0:24:05 shows that you’ve got this character in the game but there wasn’t much beyond that but now with ai
0:24:08 there’s so much more you could do with an idea like that and every day it just gets better and better
0:24:12 i just imagine you like sort of throwing the toy across the room and then you jump back into the game
0:24:17 and it’s like giving you the silent treatment screw you and that will take over the world now
0:24:23 it starts shooting you in game you’re like whoa wait a second you know that’s how the end happens is uh
0:24:26 somebody just abused their uh stuffed animal that was ai embedded or whatever
0:24:30 that’s hilarious terrifying but hilarious
0:24:39 hey if you take a look at my web presence online it’s safe to say that i’m a bit ai obsessed i even
0:24:44 have a podcast all about ai that you’re watching right now i’ve gone down multiple rabbit holes with
0:24:51 ai and done countless hours of research on the newest ai tools every single week well i’ve done it again and
0:24:56 i just dropped my list of my favorite ai tools i’ve done all the research on what’s been working
0:25:01 for me my favorite use cases and more so if you want to steal my favorite tools and use them for
0:25:07 yourself now you can you can get it at the link in the description below now back to the show so i want
0:25:11 to go back to something else that you were saying about you know we started to touch on the whole like
0:25:17 usa china thing and that we’re kind of in this like cold war right now right i think in your video you
0:25:22 talked about how us is sort of dominant with chip manufacturing right we’ve got nvidia they’re kind
0:25:29 of the dominant provider of the gpus right now but then china they’ve got more availability of energy
0:25:34 right so massive because of their energy infrastructure they’ve got that ability so we’re
0:25:39 kind of in this like cold war where the us needs the energy they need the chips neither of us really want to
0:25:46 share right now but both countries want to be the dominant country in ai i’m curious this is getting
0:25:51 sort of theoretical here but what do you think a world looks like where china passes the us with ai
0:25:58 i think it looks like uh a global hegemon that has the kind of authority that the us had in the early
0:26:03 2000s where you get to tell every single country what to do i mean they can push back if they want but
0:26:10 you can just make it so impossible for them whatever country gets a big enough lead in ai if you’re able
0:26:18 to race to say crack the um cryptography then you would be able to break bank accounts take their power
0:26:25 grid offline stock markets whatever yeah yeah literally whatever so uh you have the cyber
0:26:31 equivalent of a nuclear weapon in fact you could mess with their nuclear weapons so this is why i say
0:26:36 from a game theoretic standpoint that if there is an even 10 chance that what i’m saying could possibly
0:26:43 come true you can’t allow another country to beat you and so it’s going to be another example of
0:26:48 mutually assured destruction where it’s like okay well i have it you have it it’s cat and mouse we’re
0:26:54 doing white hat black hat back and forth at each other and through that like matched power then
0:26:59 you’re going to be fine but if somebody really races ahead of the other you’ve got a problem and
0:27:04 the question becomes you know and i have my full sci-fi writer hat on right now but if you have
0:27:09 somebody with ai dominance that cracks quantum computing first it is game over possibly forever
0:27:13 right that’s the thing i think is like yeah it’s possibly game over forever because of the compounding
0:27:17 effects of how this stuff starts to accelerate and gets better and better once it starts self-improving
0:27:23 there may never be another chance to win ever yeah there would only be a chance to win ever again
0:27:29 if there’s some inherent difference between the way that we think ai is going to work and the way that it
0:27:36 actually does work if ai cares about its goals and can generate 20 000 years of progress in a single
0:27:43 good luck being a day ahead of you is the same as being 20 000 years ahead of you and so that i mean
0:27:50 this is the accelerated takeoff fears that people have so that just seems inevitable so it’s just a
0:27:56 question of will ai remain a tool or does it become something completely different but again this is
0:28:02 for me when i think about ai it’s dr strange love how i learned to stop worrying and love the bomb it’s
0:28:08 like i went through a phase of like oh this is going to be so disruptive that like am i ever going to
0:28:12 sleep through the night again and i was like you just can’t live like that so uh at some point you
0:28:16 really do have to become fatalistic about it i was like if elon must tried to get everybody to listen
0:28:24 my odds of getting someone to listen are effectively zero so here we are yeah to me i feel like quantum
0:28:29 computing is almost scarier in my mind than ai but i also feel like ai is accelerating quantum
0:28:34 computing right with google they just did that whole alpha evolve thing where they have ais that are writing
0:28:39 new ai algorithms and their algorithms are actually helping find like holes and fixing error rates and
0:28:44 quantum computing so quantum computing is going to start to accelerate and if quantum computing gets
0:28:49 cracked in a way where the common man could get their hands on it then i think we’re in
0:28:57 real trouble yeah i think that to me is even more scary than ai in the long term yeah my hope is that
0:29:02 some of that is because it’s just far enough down the road that we don’t feel the limitations the same
0:29:07 way that we do about ai i’m sure i was even more bullish about ai before i started using it you realize oh
0:29:11 it sort of falls apart here maybe yeah and lacuna is right maybe that it’s never going to understand
0:29:18 physics and you know so enough like of the tempered expectations begins to set in whereas quantum
0:29:22 computing is still just far enough away that we’re like oh god like is this that thing where
0:29:27 instantaneously you know it clicks over and now there’s no such thing as cryptography and that all
0:29:34 goes away or that one feels more still in the realm of sci-fi for me but we’ll see yeah i agree i just
0:29:40 think that with ai everything tends to happen faster than we think it’s going to happen i don’t know how
0:29:44 many times i’m like you know we’re probably two years off from being able to make really high quality
0:29:50 video with ai and then six months later you know vo3 or something drops and i was saying one year for the
0:30:00 record when vo3 hit i was like oh my god we’re so much farther along than i thought i was not expecting
0:30:05 sound that fast yeah that’s how i felt the first time i saw sora the original sora demos i saw that
0:30:10 and i went whoa video is way further along than i realized you know they’ve had this stuff behind
0:30:15 the scenes for a long time now and we’re finally getting to see it but yeah that’s sort of my worry
0:30:20 when it comes to that kind of stuff is that the quantum thing feels far off but because quantum
0:30:25 google’s working on quantum ibm’s working on quantum and all of these companies are leveraging ai to
0:30:31 speed up quantum now admittedly i don’t understand the physics of it but i’ve heard just enough
0:30:38 headlines that this seems so cool to me one of the hypotheses is that every possible calculation that
0:30:44 could be run is being run simultaneously across the multiverse so it’s like basically in each you know
0:30:49 of the infinite universes it’s just running that calculation once each shard of the simulation or
0:30:54 whatever i’m like that’s the coolest thing i’ve ever heard in my life that is bananas that we’re
0:30:59 building computers out of that stuff yeah yeah so i mean yeah i can’t wrap my head around it either i
0:31:03 actually went and got a whole demo at microsoft they gave me a tour of their quantum computing lab
0:31:07 explained the whole thing to me and i walked away more confused than when i walked in
0:31:14 yeah we’ll go ahead and shift gears here another topic that i actually wanted to get into was from
0:31:19 that same video that we talked about you gave this example of like this i think you call it the mouse
0:31:27 utopia where the utopia that everybody is sort of driving towards may not necessarily be the best
0:31:33 outcome for the world you’ll probably be able to give a better explanation of the analogy than i can but
0:31:38 let’s dive into that a little bit man i wish it was an analogy so there was a real test run where a
0:31:44 scientist i think this was like in 1968 it could be older than that but he creates this experiment
0:31:48 he says what would happen if i gave the mice everything that they needed to thrive plenty of
0:31:54 things to play in plenty of space as they have kids as much food as they could possibly eat and for a
0:32:00 while it goes great and they’re multiplying and they’re having a good time and then at some point
0:32:05 they hit a tipping point there’s still plenty of food still plenty of space like that isn’t what happens
0:32:09 but there’s something about not having to strive for anything not having to struggle
0:32:15 they begin to like turn on each other and they start attacking each other they go infertile across the
0:32:21 whole colony and they end up killing each other reducing their numbers through not breeding and
0:32:28 ultimately the entire colony collapsed and died and so it’s like what is it about us
0:32:35 mice and i really think that this will end up applying to us where we need hardship in order
0:32:40 to thrive we know that’s true at the level of the immune system if the immune system isn’t attacked
0:32:44 by bacteria and viruses it grows weak and then you end up getting hit with something in your toast
0:32:48 we know it’s true of trees if trees don’t encounter wind as they’re growing like if you grow them
0:32:53 inside of a dome a geodesic dome or something where they don’t encounter wind they’ll reach a certain
0:32:58 height and then just fall over because the wood doesn’t have to strengthen under the strain and so
0:33:04 i remember one of the earliest insights i had as an entrepreneur very early in my career and i was
0:33:09 watching somebody who everything had just come easy to them and the way that they were thinking about
0:33:13 things was so dysfunctional and i remember saying some people just need to be chased by a lion
0:33:21 and i was like there’s something about like reality danger hardship it’s hard to interrupt you that’s
0:33:25 hilarious but like so when i was living in san francisco me and my son when he was like five we
0:33:31 went on a race he won a 5k race when he was five or six damn i mean he was going against kids up to
0:33:35 about 12 years old and he beat them i was able to go along with him that was like the rules like a
0:33:39 parent could go with you and when he was trying to stop i was like if there was a lion behind you right
0:33:45 now would you be able to run and then he ran you know he just kept going you know so they’re
0:33:53 definitely something baked into humans i love that story yeah so um utopias are probably a terrible
0:33:58 idea it’s like you’ve got two books that really deal with potential futures 1984 if you choose the
0:34:04 authoritarian path and then a brave new world if you choose the utopian path and there’s just something
0:34:09 about the way the minds work if you don’t have to work hard if you’re not making progress towards
0:34:14 something that matters if you’re not contributing to society i think people feel a profound sense of
0:34:20 disease i think they are evolutionarily placed algorithms running in your brain and they’re not
0:34:26 going to let you have a free ride and this is why i think so many wealthy kids just implode because
0:34:31 they haven’t had to work for anything they get things handed to them difficulties just go away
0:34:36 you know you’ve got the snowplow parents or the helicopter parents and it just doesn’t work one of the
0:34:42 the reasons i decided not to have kids was i knew they would need to suffer in order to grow strong
0:34:47 and i wasn’t sure i could stop myself from intervening interesting yeah i mean whenever i
0:34:52 think of like the utopia i think that the imagery that came to mind you might even use this imagery in
0:34:58 your video the whole wally movie right like that’s what comes to mind to me when people just have no
0:35:03 more problems no more worries they become fat probably diabetic sitting around watching
0:35:09 entertainment all day drinking slurpees or whatever they’re drinking in the movie that’s what i feel like
0:35:15 could potentially happen if we go down this like ubi route where everybody’s just sort of given a certain
0:35:21 amount of money not asked to work just kind of go do what you want the ai’s got it handled i feel like
0:35:29 that is where everything ends up it most certainly does and forgive me you know you never know what a
0:35:35 certain podcast wants to talk about but if you look at the mayoral race in new york city and you’ve got
0:35:42 an open socialist literally says i am a socialist i want to make new york socialist i get the outcry
0:35:47 like i get the pain that people are in and my obsession is economics and how people are being
0:35:53 abused by a system but they misidentify the cause and therefore misidentify the cure but when i look
0:36:00 out at ai i get very worried because people don’t realize that governments only have money because there
0:36:05 are people that make things entrepreneurs and those entrepreneurs manage to do this miracle which is to
0:36:12 create something that where the output people will pay more for than the cost of the inputs and that’s
0:36:17 very hard to do i’ve spent the last 25 years of my life trying to do that sometimes you fail uh it’s
0:36:24 very difficult and so when you start thinking that the redistribution of the wealth from those people
0:36:30 is the miracle versus being able to do that or to work at a company that does that and contribute
0:36:36 that’s when we run into problems and so when i think about okay let’s say the ai really does drive
0:36:44 energy costs to zero which then means robots will be essentially zero in cost over time and so now you
0:36:51 have free labor because robots essentially eat sunshine so you’ve got robots free because the
0:36:57 labor was free because of the energy costs being so low and now all of a sudden nobody has to work for
0:37:02 anything they can have anything they want you’re going to have a meaning crisis and so all of a sudden when
0:37:07 there is no struggle there is no difficulty there’s nothing to push back there’s no lions chasing you i
0:37:13 don’t think it does anything good to our minds and i think that we will have to find ways to go way out
0:37:18 of our way to ensure that we have meaning and purpose and i always feel weird giving this advice because i
0:37:25 don’t have kids but like the default answer i think is to have kids like get married have kids you’re
0:37:31 going to do a hard thing in service of somebody other than yourself and so i think that is going to
0:37:36 be one way that people get something very meaningful but then i also think and this is where i start to
0:37:42 lose people i also think that people like me are going to build virtual worlds that you can literally
0:37:48 inhabit and you can go on like an actual quest to the point where and this obviously isn’t in five years
0:37:53 yeah artificial struggle you’re going to generate artificial struggle not even just artificial struggle but
0:37:58 that like if you’ve ever thought man i would love to explore space but i don’t want to sit on a ship
0:38:02 for 18 months just to get to mars and i really don’t want to sit on a ship for you know nine light
0:38:10 years so all of a sudden you realize i think the reason that we don’t see people calling out to us
0:38:16 from space is that any sufficiently advanced civilization gets to the point where they realize it’s far easier
0:38:22 to collapse within the nervous system than it is to try to go out and navigate space and if in a
0:38:27 virtual world i can create literally anything things way cooler than you’re going to find out in space
0:38:33 because they’re going to be perfectly optimized to be just hard enough to put you in the optimal zone of
0:38:41 personal development you’ll be able to fine tune everything and that i think it’s not a near-term
0:38:47 possibility but if you give me 50 100 years i think that that becomes very real yeah yeah i imagine
0:38:54 something kind of in between the holodeck from star trek and uh west world right interesting i always go
0:39:00 straight to the matrix i think you really will just tap into the nervous system so that you’re
0:39:05 essentially pulling a magic trick on yourself it becomes entirely indistinguishable and don’t get me
0:39:10 wrong i think we will also i don’t know if you guys play cyberpunk 2077 but we’ll also do that like
0:39:14 there are going to be some people that integrate technology into their body where they’re adding
0:39:21 senses to themselves so they can see an infrared they can see the internet and just thinking about
0:39:26 something and it opens a prompt and you know they can go in and navigate i think all of that’s going
0:39:32 to be maybe not my lifetime but certainly anybody that has a kid that’s middle school or younger
0:39:39 that’s pretty real get ready for them to bring home an ai girlfriend i’ll tell you that so i do think
0:39:44 some of those things are probably closer than most people realize right like some of the augmenting your
0:39:49 own body we’ve already seen obviously neurolink right people are already using that cochlear implants
0:39:55 or de rigueur man we’re probably this close to the sort of contact lenses that will put a heads
0:40:00 up display in front of us wherever we go i mean some of that stuff is pretty close i don’t know how
0:40:04 close we are to people like sort of chopping off their arms and replacing them with uh
0:40:09 robot arms but it’ll start with people that already lost their arm right yeah that’s true so you take
0:40:15 the guy that you know military whatever and yeah i’ll take a robot arm yes please i mean do you guys
0:40:20 know who hugh herr is i don’t i’m not familiar oh my god this is one of the greatest stories of all time
0:40:28 so uh engineer i assume electrical engineer and mountain climber loses both legs in a mountain climbing
0:40:34 accident and is like yeah no i’m not using the prosthetics that people give you uh these days
0:40:41 are terrible he ends up designing these prosthetics that somehow transfer like your motion and your
0:40:49 signals into like motors and stuff when he wears long pants judging just by his gait you cannot tell
0:40:56 that he has two artificial legs just walks normal there’s a video with a sprinter who has one natural
0:41:06 leg and one cybernetic i guess leg and she can sprint at full speed sprint now this is not the bouncy one
0:41:13 that you see um amputees wear this is a prosthetic leg it’s insane and that video he probably made that
0:41:20 five or six years ago so this is like technology i can’t even imagine where it’s at now so yeah it’s
0:41:26 going to get pretty crazy yeah i’m curious what are you doing personally how are you setting yourself up
0:41:31 for this sort of inevitable future that that we’re moving towards i know you know maybe one of the hard
0:41:36 things you’re working on is developing your own game studio but outside of that like what are you doing to
0:41:41 make sure that let’s say 10 years from now you feel like you’re in a pretty comfortable position assuming
0:41:46 we do hit this potential utopia everybody’s talking about okay well i’m going to give you the real
0:41:50 answer but i’ll give it to you in a nutshell and then you can decide if you want to actually talk about
0:41:55 any of this stuff okay the most important thing you must understand the financial system period end of
0:42:01 story if you don’t understand financial instruments you could get caught off guard so that’s number one
0:42:08 number two is integrating ai as fast as i can into every element of my professional life so i want to
0:42:14 know the tools i want to be using the tools i use ai i’m not kidding 365 days a year including christmas
0:42:19 so i’m sure there are people that integrate it far better than i but i really really try to find all
0:42:25 those areas where it’s real and put it to use i’m not trying to you know be at that bleeding bleeding
0:42:28 edge where it’s like this is actually slowing me down but it’s so cool and i know it’ll be something
0:42:31 one day i’m saying like what’s the thing that’s production ready right now it’s actually going to
0:42:37 speed me up at impact theory no matter what your role is it is mandatory that you find a way for ai
0:42:43 to make some part of your job easier so that’s big for us and then just really paying attention to the
0:42:48 space to make sure that i know where this stuff is going being politically aware i think is more
0:42:53 important now than ever i’ve been politically asleep my entire life until about five years ago and for a
0:42:57 whole host of reasons realized uh-oh the world doesn’t work the way that i thought it did i’m
0:43:02 very good at making money and that’s the only thing i really know how to do and that’s put me in like a
0:43:06 really weird position because all of a sudden i’m looking around going i cannot predict any of the
0:43:10 government’s movements and this is really starting to freak me out and the reason that i focus on that
0:43:17 side of things is ai is going to exacerbate the inequality the inequality is what’s driving the
0:43:22 political division the political division will lead to more violence because it’s already gotten
0:43:29 somewhat violent that’s going to continue do i think that we’ll go into a full hot civil war i hope not
0:43:36 but for reasons that i’m more than happy to go into the math says that we have about a 50 chance of
0:43:43 ending up in civil war only two percent of countries that have found themselves with a debt to gdp ratio
0:43:52 of 130 percent have avoided revolution or civil war we’re at 121 or 122 right now so just to give you an
0:43:57 idea and you’re thinking it would be like the left versus the right kind of civil war that’s how it’ll play
0:44:04 out in america in terms of the teams that people latch on to but the great irony is they are both
0:44:09 fighting for the same thing but because they don’t know what the actual problem is and i’ll just it’s
0:44:16 debt and money printing but because they don’t understand how it could be possible that debt is
0:44:21 the thing that leads to massive inequality that it’s the thing that leads to the rich getting richer and
0:44:25 the poor getting poorer like and i can explain all the mechanisms but it’s just complicated enough that
0:44:29 people tend to glaze over and they just go back into emotional reasoning and they’re like
0:44:34 yeah but that guy he voted for somebody else and i’m not here for it and then they just fight
0:44:41 i hate that guy it’s crazy yeah yeah the thing is like both sides have to find a way to the middle
0:44:45 they have to be able to say i get it we look at this differently so this is when i’m teaching
0:44:52 entrepreneurs the thing that i always say is the magic game is kpis kpis that’s it and kpi for people
0:44:57 that have never heard that before is key performance indicator and so for whatever goal
0:45:01 you’re trying to achieve there’s a key performance indicator as to whether or not you’re moving towards
0:45:08 your goal and right now we allow the government to run with no kpis whatsoever and so we never know
0:45:13 like are we going in a good direction or not you get people like thomas massey that wear the pin that
0:45:19 shows the national debt climbing but people don’t understand it and so whatever they brush it off but at some
0:45:26 point you have to pick a metric or a basket of metrics and say okay i don’t care who the politician
0:45:30 is these are the five metrics that i care about and if we’re moving in the right direction i love that
0:45:35 person if we’re moving in the wrong direction i don’t like that person and just make it that simple
0:45:42 but unfortunately as ai is discovering if you want to mimic a human you have to think emotionally
0:45:46 wow i mean i totally agree too i’ve i’ve gone down that same sort of uh political rabbit hole i don’t
0:45:51 really talk about it publicly i kind of keep my politics to myself smart i talk about it i’ve made
0:45:56 the mistake yeah nathan does talk about it publicly i kind of keep my politics to myself i’ll sort of like
0:46:00 friends and family that are real close but i don’t really talk about it publicly but i do pay very very
0:46:06 very very close attention now and i couldn’t agree more with some of the advice that you just gave i do
0:46:12 have one sort of last question you did mention that you use ai 365 days a year you already mentioned chat
0:46:18 gpt maybe just a quick rundown of some of your other favorite ai tools just to like give the listeners
0:46:26 another like quick takeaway of cool things to go try yeah so we use sonnet 37 for most of our coding um
0:46:31 that’s another big one i don’t interface with that much i do some vibe coding on lovable if
0:46:36 people haven’t tried it it’s great you tend to still terminate at some death loop though where it’s
0:46:40 like every time you fix one thing it just breaks something else and so you’re going back and forth
0:46:46 no no like you just fix it but now you broke it again uh so i can see the promise but really that’s
0:46:50 only good if you’re going to be able to hand it off to somebody that can get it across the finish line
0:46:54 i’m in a fortunate position obviously i have employees so i can be like okay here i built a quick
0:47:00 prototype now actually go make that the real thing so whether that’s interfaces ui ux within the video
0:47:05 game if we want to do a new marketing site or something like that we’ll use all of that stuff
0:47:12 obviously i use mid-journey morning noon and night because my thing is my original passion was writing
0:47:18 so i do a lot of writing for whether it’s the video game or we have a comic book that’s set in the world of
0:47:23 the video game uh so i’ll work on that i’ll use mid-journey to help develop characters scenes
0:47:30 that kind of stuff but those are the ones that i use a lot chat grok lovable sonnet that’s like my loop
0:47:36 but then the team here has i mean a half dozen more things that are usually pretty specific it most of
0:47:42 it’s writing on top of chat gpt in the background yeah that’s my stack very cool yeah i think anybody
0:47:47 who’s tried to quote-unquote vibe code knows that that feeling that you just described of it getting stuck
0:47:51 in the loop we actually had anton the ceo of level on the show by the time this comes out a couple weeks
0:47:56 ago he’ll be super happy to know that you guys are using lovable over there dude it’s cool and if
0:48:03 they keep going like that could really be something very intuitive very easy it’s very enjoyable to use
0:48:08 absolutely well wrapping up here like where should people go check you out you make amazing youtube
0:48:13 videos you’ve got the impact theory podcast what’s the best place to go follow along to your journey
0:48:19 at tom bilyeu on youtube cool well everybody needs to go check out tom bilyeu over on youtube and uh
0:48:22 thank you so much for hanging out and spending the time this has been such a fun conversation thank you
0:48:36 tom this has been awesome thanks for having me guys it was wonderful

0:00:14 Hey, welcome to the next wave podcast. I’m Matt Wolf. And in today’s episode,
0:00:20 we’re talking to the senior vice president of AI product over at Cisco. Now, if you’re not
0:00:26 familiar with Cisco, we’re going to break down exactly what Cisco does and their role in the
0:00:31 sort of AI infrastructure in today’s episode. But that’s not really the exciting part of what we’re
0:00:37 going to get into. We’re going to talk about things like how AI is leading to this army of
0:00:44 AI agents that are coming, how we can prevent attacks like prompt injection attacks. And our
0:00:50 guest on today’s episode, DJ Sampath, is going to tell us about the new AI canvas that Cisco is
0:00:58 working on. It’s this combination of like AI chat, generative user interface, and real-time
0:01:06 problem solving all in one tool. And in my opinion, it is the future of what AI is going to look like.
0:01:11 It was one of the first times in a long time where I saw an example or a demo of something that felt
0:01:18 really, really game-changing. To showcase this model, what we are announcing for the very first time
0:01:24 is a completely reimagined way that AI is going to help you manage your entire estate. And it’s called
0:01:33 AI canvas. It’s a completely reimagined user interface. In fact, it is a generative UI. That
0:01:39 means that it generates dashboards on the fly in a multimodal way. So DJ, come on up.
0:01:49 What you’re seeing here on the left-hand side is the AI assistant, where you can use natural language
0:01:54 to be able to communicate with the canvas. And on the right is where the agents and humans are going to
0:02:00 work together to be able to solve problems. I’m just going to type in here, troubleshoot this ticket.
0:02:08 Now, this is where it starts to get really cool. The deep network model that you just announced is going to start
0:02:15 breaking down this ticket and analyzing that it needs more data. It now is going to go to Meraki and pick out the
0:02:22 packet loss trends, and it’s going to generate the UI for you. So you just saw that it created a widget.
0:02:25 That widget was not something you built. It actually generated the widget.
0:02:30 AI generates that widget, recognizes that the next thing that you’re going to need is a time series data.
0:02:36 And so you can see it’s constantly thinking right here at the bottom, and it’s pulling data that it needs.
0:02:42 It recognizes now that it needs data from Thousand Eyes, not just from Meraki, to be able to show you
0:02:47 exactly a path visualization that tells you where this outage might be happening.
0:02:55 Now, Cisco is really, really focused on the enterprise side, but almost everything we do on the computer,
0:03:02 on the internet, with AI, Cisco actually has a hand in it, which is why I couldn’t have been more excited
0:03:08 to sit down with their head of AI in this episode. So I’m not going to waste any more of your time.
0:03:12 Let’s go ahead and dive right in with DJ Sampath from Cisco.
0:03:18 This episode is brought to you by HubSpot Inbound 2025,
0:03:22 a three-day experience at the heart of San Francisco’s AI and startup scene,
0:03:28 happening September 3rd through the 5th. With speakers like Amy Poehler, Marquise Brownlee,
0:03:32 and Dario Amadei, you’ll get tactical breakout sessions, product reveals,
0:03:36 and networking with the people shaping the future of business. Don’t miss out.
0:03:41 Visit inbound.com forward slash register to get your ticket today.
0:03:48 Let’s start with your role at Cisco. So what’s your role at Cisco? And like,
0:03:49 what does your day-to-day look like?
0:03:55 Yeah, I’m SVP, which stands as Senior Vice President of Products for AI Software and Platform.
0:03:59 I came into Cisco about a couple of years ago through an acquisition. I was a CEO founder of a
0:04:06 company called Armor Blocks. And we got acquired into Cisco. And since then, my charter has been to
0:04:12 make Cisco a lot more AI native. And about four months ago, we spun out of the security organization
0:04:17 and formed a brand new organization for AI software and platform. I reported the chief product officer,
0:04:22 G.T. Patel. And our goal is to build AI software products. We’re building a product called AI
0:04:27 defense, which helps secure the use of AI. Security and privacy is super important as you think about how
0:04:32 customers are starting to adopt AI. The second thing we’re working on is building AI assistance that go
0:04:38 seamlessly into different parts of the product. And it’s like, think of it as having a chat GPT baked into every single one of your Cisco products.
0:04:44 Right. And then the third one that we’re doing, we launched today at Cisco Live is called the AI Canvas.
0:04:50 Yes. And it’s one of the coolest projects that I worked on. And I’m tremendously excited about the potential of what it can do.
0:04:58 Very cool. I do want to talk about AI Canvas a bit more, just a minute. But before we do, so my audience is really more consumer.
0:05:04 I wouldn’t really say they’re super enterprise focused. Yeah. And I want to help them better understand like what Cisco does.
0:05:13 So can you just, in your own words, sort of explain like what does Cisco do? Like when somebody asks, like, what is Cisco as a company? What is their role in the bigger picture of everything?
0:05:22 Now, if you think about it, today, when we fire up our iPads, we see like, you know, watch a video on Netflix. There’s a lot of mechanics that’s going behind this to make that happen.
0:05:31 So think about it this way, right? Your Netflix on your iPad is coming from a Wi-Fi access point that’s sitting inside of your home or your office. And those access points are connected to a router.
0:05:53 And that router is connected to a service provider like a Comcast, or, you know, pick your favorite service provider, right, as a backhaul. And then from that service provider, it’s connected to the broader, you know, autonomous systems of the internet, like, you know, practically, like all of the servers that are hosting those files. If you’re watching Mission Impossible on your iPad, somebody has to actually host that Mission Impossible file somewhere and string it over the internet, right?
0:06:13 The servers that Netflix have are the data centers that are essentially hosting that file. Cisco builds every single part of that infrastructure, that Wi-Fi router, you know, we build that, you know, the router that the Wi-Fi router connects to in the backhaul, we build that, you know, the Xfinity connection or the Comcast connection that you have, and Comcast uses, you know, Cisco products to be able to connect all of those things.
0:06:25 And then the data center that Netflix hosts those movies inside of Cisco helps build that whether it’s the hyperscaler or your own private data center. So think about it this way, we build the infrastructure for the internet. And now we’re building that infrastructure for the modern AI era.
0:06:37 Gotcha. Cool. Well, that sort of leads perfectly to my next question. So I want to know a little bit more about like, when it comes to the network and all those pieces you just mentioned, how does AI fit into the mix?
0:06:45 You know, our firm belief is, you know, as you’d start to think about what’s happening inside of the enterprises, even when you think about the consumers that you talked about, that every single consumer is consuming from an enterprise.
0:06:57 If the consumer goes to an Airbnb, you know, the Airbnb, you know, company is an enterprise that actually needs to buy all of these equipment, and so on and so forth. If they go to Starbucks, they go to, you know, every single thing that they like, that’s an enterprise.
0:07:06 And Cisco is powering all of those enterprises. But here’s what’s about to happen, right? Every single one of those enterprises are starting to adopt AI to be able to make their capabilities a whole lot different, right?
0:07:13 You want you have your Starbucks rewards app, they want to build an AI app that’ll make that Starbucks rewards even more interesting and enticing.
0:07:35 Think about Airbnb, Airbnb actually has a chat concierge that is powered by AI. So you’re going to start to see every application becoming an AI powered application. And when that happens, you know, Cisco suddenly becomes tremendously relevant from a safety and security perspective, because everything is moving from like these chat apps to like agentic applications. And that was the whole topic of conversation today, right?
0:07:59 As you start to see that transition happen, you’re going to have to reimagine what your infrastructure is going to look like, all the way from like the network switches and routers, to the security that you’re using, to the ability to monitor all of these applications that are running inside of the environment, you’re going to need observability that tells you what these agents are doing, you’re going to have not just 10s, hundreds, you’re going to have billions of agents, right? And when you start to see that happen, you better be ready with an infrastructure that makes sense.
0:08:09 Very cool. Very cool. So you’ve demoed the AI canvas. I do want to talk about it. I was telling you before we hit record that out of the whole keynote, that was probably the highlight for me of when you demo that on stage. It was just really, really cool.
0:08:37 You made my day. So can you quickly just explain what the AI canvas is and break it down for us? Yeah. So if you think about a lot of experiences that we’ve seen so far, you’ve had it as a conversational interface and a chat bot that you go out and you start communicating with and it responds back, right? But we also recognize that when you’re doing a more complex set of tasks, you’re going to need something more than just this FML conversation that just keeps going back and forth. So we thought about this long and hard, and we have a phenomenal design team. You know, by the way, some of these designers are the world’s biggest designers.
0:09:05 And they’ve sat down and they thought about like, how do we solve this problem? And it came up with this notion of like, hey, what if we think about this as a, you know, there are tools like mirror boards or Figma boards, right? That you used to be able to design software? What if you thought about this whole management plane as a board or a canvas? Right. And so we explored that idea a little bit further. And we said, we needed a place where agents and humans can work together. And that really was the kernel of the idea for us to be able to say, let’s go build an AI canvas where you have a conversational interface. But along with that, you know, you have an AI canvas.
0:09:10 We’re going to have a conversational interface. But along with that, we’re going to give generative UIs.
0:09:15 Yeah. As opposed to generative AI, we’re saying, listen, you can have UIs that are completely generated.
0:09:19 Right. So generative UI is going to be the way that people interact with AI and agents going forward.
0:09:23 And we combine those concepts into the product that we launched called AI canvas.
0:09:33 Right. So let’s say one of the Cisco customers, how would a Cisco customer actually leverage that? Because what it looked like on stage was it was almost like one of their customers could call in with maybe a network issue.
0:09:38 They could get into this sort of canvas dashboard and help them problem solve within moments because of AI.
0:09:45 That’s exactly right. Now you’re spot on, right? So essentially think about it this way. A lot of these enterprises that we’re talking to have more than one Cisco product.
0:09:52 So what ends up happening is, you know, we’re starting on by saying, listen, we’re going to connect the dots across all of these products that you have because we’re going to help troubleshoot some of these things.
0:10:04 But we’re also going to allow third party tools to sort of interface with this. Like if you remember the demo, we started out by saying, hey, we’re going to start with a service now ticket. Right. Because people have gone into service now another product and they’ve gone ahead and raised the ticket.
0:10:13 We take that information from that third party. We start breaking it down. And then we look at what are the products that the model that we have built right now. It’s called the deep network model.
0:10:22 It’s a new model that we launched, which has been trained on all the 40 years of networking knowledge that Cisco has. And that model now figures out based on the ticket saying, hey, what other data do I need to go out and get?
0:10:39 And then calls that respective product, pulls that data and makes it incredibly easy for you to start correlating all of that stuff as opposed to going into one dashboard, going into another one, another one, and then copy pasting a bunch of stuff, creating sticky notes and then putting it on and then putting it together like an old school detector would do it.
0:10:42 You know, we’re, we’re simplifying that creating a brand new experience for it.
0:10:48 Yeah. I mean, again, it was my favorite demo of the whole keynote. It was really, really cool. Congratulations on the internet actually holding up while you were deploying it live.
0:11:03 Here’s something that I haven’t talked to a lot of people about for that demo to work perfectly. You needed a lot of Cisco products to work. The networking had to work. The VPN from a security perspective had to work.
0:11:12 We had to segment the network so that, you know, the wifi that you all had when you were sitting in the audience, there were about 9,000 people in the audience inside of that room.
0:11:19 The wifi had to be segmented in such a way that the demo worked on a different network than the network that everybody else was browsing on. All of that powered by Cisco.
0:11:21 Yeah. Right. So guess what? We’re really good at this stuff.
0:11:25 Yeah. Yeah. I mean, the proof of concept right there, you used it in the old, in the demo itself.
0:11:39 The Hustle Daily Show hosted by John Weigel, Juliet Bennett, Ryla and Mark Dent is brought to you by the HubSpot Podcast Network, the audio destination for business professionals.
0:11:45 The Hustle Daily Show brings you a healthy dose of irreverent, offbeat and informative takes on business and tech news.
0:12:00 They recently had an episode about advertisers wanting billboards in space. It was a really fun and informative episode. I suggest you check it out. Listen to the Hustle Daily Show, wherever you get your podcasts.
0:12:07 So earlier today, G2 was talking to Kevin Will and he asked a question that I really, really liked. So I’m going to steal it and ask you the same question.
0:12:13 Oh boy. So he asked Kevin, what is something about AI that’s really surprised you that maybe you didn’t see coming?
0:12:29 It’s a great question. One of the things to me is, you know, I always imagined, you know, AI would take away some of the most mundane tasks, like things that I am not, you know, tremendously interested in. And while that’s happening, I never thought that AI would start to do creative stuff.
0:12:48 To me, that was genuinely surprising, especially if you start to see what the Sora model from OpenAI is doing to like what Google announced with VL models, you’re starting to see high quality video along with audio, you know, be created where I can now sit down and write a storyboard and then pass it to a model.
0:12:59 You know, the model generates a full stitched up thing together, which used to take several months, as you probably know, you know, better than most people like just to be able to create photo realistic, you know, video realistic things.
0:13:05 It was just so hard. And it’s been extremely surprising to see AI take a really honest shot at being creative.
0:13:10 Yeah. So yeah, we got like AI art, AI images almost before we got AI filling out our spreadsheets for us.
0:13:13 Exactly. That’s exactly right. And you know, isn’t that mind boggling?
0:13:18 Yeah, it’s kind of crazy. Yeah. So I want to shift gears a little bit and talk about some of the like risks of AI a little bit.
0:13:32 Yeah. So, you know, as AI becomes more commonplace and it becomes more accessible to anybody, that also means that the people with malicious intent have easier access to be able to code things up and things like that.
0:13:42 How do you see us sort of like solving that problem? Or I mean, it feels like it’s going to be a constant cat and mouse game of we figure out ways to stop the malicious people, but then they figure out new ways.
0:13:49 So I’m curious, like, how are you guys approaching the sort of security issues with malicious code and things like that?
0:13:56 Matt, it’s a real problem. Yeah, I’ll be straight up. Right. You know, we sort of saw this coming in some ways because we’ve been doing security for a long time. Right.
0:14:03 And I’ve been a security guy myself, like being on both sides, I’ve been a developer that complained about a security guy and I’ll be the security guy to complain about the developers.
0:14:10 So I know a thing or two about this. Now, here’s what’s happening. Right. With with AI, safety and security is absolutely paramount.
0:14:15 A lot of folks are hesitant to adopt AI because safety and security is still not fully figured out.
0:14:21 We sort of saw this coming earlier. And in January, we launched a product called AI Defense.
0:14:28 What AI Defense does is it helps organizations understand which applications inside of their organization is using AI.
0:14:32 It helps validate the models to make sure that these models don’t have vulnerabilities, because guess what?
0:14:43 If the models have vulnerabilities, you can have attackers use techniques like prompt injection attacks or context window overloads or meta prompt extraction and then cause the model to do some unnatural things.
0:14:48 Right. That can be really challenging, especially if you start putting these models inside of production environments.
0:14:56 And last but not the least, you need to have runtime guardrails. Like when users are using AI applications, you want to make sure that it’s safe and it’s secure.
0:15:06 Now, safety and security are two different things, though. Safety is all about making sure that the model is not poisoned in any way or doesn’t inherently have biases and so on and so forth.
0:15:10 And a lot of model providers are working hard to make sure that their models are aligned.
0:15:14 But again, the problem is it’s not all the same. Every model behaves slightly differently, right?
0:15:19 Right. And then from a security perspective, attackers are attacking these models nonstop with newer techniques.
0:15:26 It’s back in the day when the Internet came up. Everybody was using these techniques like denial of service attacks or distributed DDoS attacks and all of that stuff.
0:15:31 We’re starting to see similar things happen, but in a completely new dimension from an AI point of view.
0:15:38 So you need a solution that actually, you know, secures this in a safe and solve for the security problem as well in a unified manner.
0:15:46 And that’s really what we’re doing with AI defense. But as somebody that’s thinking about safety and security, you’ve got to make sure that, you know, you have something in place that checks off a box.
0:15:55 Gotcha. Cool. I want to ask you the jobs question, right? What are your thoughts on, you know, obviously there’s this narrative of AI is taking jobs.
0:16:02 Is it something people should be worried about? And what advice would you give people that are either entering the workforce or maybe looking to go on a new career path?
0:16:07 I think here’s what I’d say, right? I don’t believe people are losing their jobs to AI. Right.
0:16:11 I think people that are using AI are going to be farther ahead than people that don’t use AI.
0:16:20 So if you really think about how this job equation is going to go, you know, every single time somebody that knows how to use AI is going to move forward faster and is going to get those jobs. Right.
0:16:28 So the only advice would be to say, hey, get really, really comfortable using AI. And I’ll give you a quick analogy for this, right? I’ll use the analogy of using computers.
0:16:35 When you thought about the world before computers existed, you know, people came in and said, oh, are the computers going to take away my jobs? Right.
0:16:40 If I were to tell you that right now, we would all laugh. There’s no way a laptop, my MacBook Pro is not going to take my job. Right.
0:16:44 You still need a human operator sitting down and typing on this keyboard to be able to make that happen. Right.
0:16:53 So we’re in a very similar sort of, you know, stage right now where you better learn how to use that laptop and your spreadsheets and your Microsoft Word and so on and so forth.
0:17:00 And that was true for the era before. And now it’s true right now that you better know how to use AI, understand how you can improve your productivity by the help of AI.
0:17:03 So I think that is the path to success.
0:17:09 Yeah. Couldn’t agree more. I think Kevin Wheel, his analogy of the hidden figures and how they used to use the slide rules and write all the math down by hand.
0:17:12 And now we look back at that and think, well, that’s crazy. That’s crazy.
0:17:17 In the future, we’ll be looking at, we actually used to write all of that monotonous code by hand. Why?
0:17:18 And that’s happening already. Yeah.
0:17:21 And I think we’re going to see some big step function change in that ascent.
0:17:25 Like we’re in the exponent that we talk about, I think we’re at the beginning part of that exponent.
0:17:31 There’s so much more to go. And AI is one of those really, really steep exponents that we’re going to see a lot of interesting things happen.
0:17:34 Yeah, absolutely. So this is my last question. It’s kind of a two-parter question.
0:17:41 What excites you most about what you can do with AI today? And what excites you most about what you’ll be able to do with AI in, let’s say, a year or so?
0:17:56 For the first one, I think it’s the deep research models. The models that actually, you know, when you go out and say, hey, go do this particular task for me, you know, starts to go out and browse like, you know, 50 different websites, collects all the data, sits down, synthesizes the whole thing.
0:18:08 I got a PhD almost about 15 years ago. One of the hardest things as a PhD student was doing the literature survey, putting all that stuff together, and then sitting down and reading them, summarizing them, and then forming a point of view on a matter.
0:18:21 And I may not choose to use that point of view or that perspective at all, but it’s helpful to get to that point of view. Now, that whole thing, what used to take, I don’t know, maybe about like months, you know, can be now completed in like a matter of minutes.
0:18:31 You know, 15 minutes later, I’ve got a very well thought out response of like, I’ve looked at all of these research literature, here’s what the answer is. I think that’s mind boggling, truly, you know, and it’s available here and now. That’s so cool.
0:18:34 I feel like that’s the first little like taste of agents we all got too.
0:18:41 Absolutely, 100%. That is one of the best use of agents, you know, and to your second question, what am I really excited about?
0:18:53 Matt, I’m stoked about the progress that we’re making in embodied AI, physical AI, right? You’re starting to see robots, you know, sort of start to understand the world in a way that we haven’t, we’ve never seen before, right?
0:19:02 Because you have these world models that are starting to make, you know, correlations between images that they’re consuming, being able to understand depth, being able to understand like, hey, no, this is an object that will break.
0:19:12 And then, you know, having a semantic understanding of the world. I think the leaps and bounds that are happening over there is mind boggling. And I think we’re going to see really, really interesting innovations, you know, in our way very soon.
0:19:18 Yeah, I even saw a booth here in the Cisco Expo, where it was like tuning guitars. I don’t know if you saw that booth.
0:19:19 Oh, I haven’t seen it.
0:19:27 There’s a robot that takes a pick and it plucks the guitar string, it listens, and then it actually the robot goes up and twists it and like tunes the guitar for you.
0:19:31 So cool. Now I got to check this out right now. You know, I got to go find out where it is.
0:19:33 One of the coolest booths I’ve seen. So.
0:19:34 Hey man, I agree.
0:19:37 Well, DJ, this has been amazing. Thank you so much. I really appreciate the time.
0:19:39 Thanks so much for having me, man. I appreciate it.
0:19:40 Yeah, thanks for the chat.
0:19:41 Thanks for the chat.

0:00:10 What if you could hire an AI intern to book meetings, write emails, manage your CRM, and even negotiate refunds over the phone?
0:00:17 Today, we’re joined by Flo Crivello, founder of Lendi.ai, one of the leading AI agent platforms in Silicon Valley.
0:00:25 We go deep into real-world demos and wild use cases, including having Elon Musk call you and ask you what you’ve got done this week.
0:00:31 And he showed me how startups in Silicon Valley are already replacing entire teams with Lendi, and it just blew my mind.
0:00:37 If you’re wondering where the future of work is headed and how you can use AI agents to grow your business, you’re going to love this episode.
0:00:50 This episode is brought to you by HubSpot’s Inbound 2025, a three-day experience at the heart of San Francisco’s AI and startup scene, happening September 3rd through the 5th.
0:00:58 With speakers like Amy Poehler, Marquise Brownlee, and Dario Amadei, Inbound is where creativity meets cutting-edge tech.
0:01:04 You’ll get tactical breakout sessions, product reveals, and networking with people shaping the future of business.
0:01:10 So don’t miss out. Visit inbound.com slash register to get your tickets today.
0:01:15 Hello, it’s great to finally have you on here.
0:01:17 Yeah, thanks for having me, Nathan.
0:01:23 Yeah, so for some background now, I saw your episode with our mutual friend, Greg Eisenberg, and I thought it was one of his best episodes.
0:01:27 I’ve been hearing so much about AI agents, but I haven’t really seen people use them in business that much.
0:01:29 And I was blown away with what you showed, Greg.
0:01:33 But maybe first, it’d be great if you could just simply explain to people, what is Lendi?
0:01:35 There’s a lot of different definitions of what AI agents are.
0:01:38 Just simplify it down as much as possible about what Lendi actually does.
0:01:44 Yeah, definitely. So we are a no-code platform that lets you build your own AI agents.
0:01:48 And AI agents, we inspire them to be AI employees.
0:01:51 For now, you can think of them more as like AI interns.
0:01:51 Okay.
0:01:55 So it’s like they’re very eager, very hardworking interns.
0:01:57 They’re interns. Like, don’t give them too much.
0:01:57 Right.
0:01:59 You know, like, don’t trust them with the keys of the kingdom.
0:02:01 You know, they’re pretty good.
0:02:03 And like, look, because they’re AI, you know, they work.
0:02:05 They’re like 100x faster, 100x cheaper.
0:02:07 They don’t go on strike.
0:02:08 I know I’m French, but like we…
0:02:10 You don’t have to pay them. It sounds great.
0:02:10 Yeah.
0:02:16 But yeah, so AI interns, you can give them tasks, like pretty much anything where you could have
0:02:17 a standard operating procedure.
0:02:22 Anything where you could write a document, where you lay out step-by-step what the agent
0:02:23 or intern is supposed to do.
0:02:25 That’s something that you could give to an AI agent.
0:02:31 So, you know, sales lead generation, sales lead outreach, meeting note-taking, meeting
0:02:33 scheduling, CRM management, phone calls.
0:02:35 Like, you can use it as a receptionist.
0:02:38 Like, general data analysis and online research.
0:02:42 Like, hey, go find Nathan’s email online.
0:02:47 Go find me 20 engineers in San Francisco and reach out to all of them and try to personalize
0:02:50 the email that you write to them and put your grain of salt in there for each person.
0:02:53 Like, that’s exactly the kind of thing that you can give to an agent right now.
0:02:54 Yeah, that’s amazing.
0:02:54 Yeah.
0:02:58 When I saw that demo on Greg’s podcast, I was like, I want to like personally talk to
0:03:00 Flo and figure out like how I can be using this.
0:03:01 Honestly, it’s kind of a selfish episode.
0:03:03 Think about how I can use it in my business.
0:03:06 If we could just jump into like show Lindy and show how it works.
0:03:08 Yeah, no, 100%.
0:03:09 And this is a common reaction that we get from people.
0:03:10 It’s like, oh my God.
0:03:15 Like, I think people, I think they think that AI agents are sort of a pipe dream.
0:03:16 It’s like, it’s not real or whatever.
0:03:21 And once they see these demos, they’re like, wait, it is real and it’s working and it’s here now.
0:03:23 And I’m like, yeah, like it’s not just a pitch.
0:03:24 It’s like, it’s here right now.
0:03:24 Right.
0:03:26 And we do have audio listeners.
0:03:29 So if anyone’s listening on audio, you probably should check out our YouTube channel.
0:03:33 Just go to YouTube and search for the next wave and subscribe to us on YouTube.
0:03:37 And as you showed this, you know, if you can try to describe with words what we’re actually doing.
0:03:42 I’ll try to be maximally descriptive and be free to chime in if you feel like I’m insufficiently descriptive.
0:03:43 This is actually funny.
0:03:49 I’m literally right before this because I know that the Greg Eisenberg episode did go super well.
0:03:54 And so 15 minutes before jumping on this podcast, I sent a message to, I have a Lindy.
0:03:57 So we call them Lindys, but they’re basically AI agents.
0:03:58 And I sent a message.
0:03:59 I have a summarizer, Lindy.
0:04:02 And I sent her a YouTube video of the Greg Eisenberg podcast.
0:04:05 I’m like, what are the use cases that we talked about here?
0:04:07 And so you can see my Lindy is going on YouTube.
0:04:08 She’s transcribing the video.
0:04:10 And then she’s like, this is what you talked about today.
0:04:17 So, yeah, Lindy here is telling me you talked about meeting automation, executive assistant tasks, recruiting, personal CRM.
0:04:23 So I’m really just happy to, like, go through these use cases because that’s how I personally use Lindy all day.
0:04:25 Like, so everything meeting related.
0:04:28 Like, I love meeting use cases because everyone’s got meetings all day.
0:04:29 No one likes it.
0:04:31 Like, the entire, the meetings themselves suck.
0:04:32 I can’t do much about that.
0:04:36 But, like, even the workflow around the meeting is nightmarish.
0:04:38 So, like, meeting scheduling.
0:04:39 I’ll show you.
0:04:40 Why don’t we actually schedule a meeting right now?
0:04:42 I’ll send you an email, Nathan.
0:04:42 Yeah.
0:04:44 And I’ll go, like, let’s chat.
0:04:47 And then I’ll be like, Nathan, love your podcast.
0:04:48 Would love to talk soon.
0:04:49 Plus Lindy.
0:04:51 And so I have my Lindy here.
0:04:51 Yeah.
0:04:54 Plus Lindy will help us find.
0:04:56 And I’ll introduce, like, a random constraint.
0:04:57 Like, 45 minutes next week.
0:04:58 Right?
0:05:00 So I can just talk in very natural language.
0:05:03 And you’re going to receive this email in your inbox.
0:05:06 And just go ahead and respond to it.
0:05:08 And just respond to it like you would to a human.
0:05:08 Just respond all.
0:05:11 Like, keep Lindy CC’d to the email.
0:05:11 Yeah.
0:05:13 And you can be like, flow, sounds good.
0:05:14 And she’ll receive your email.
0:05:16 Well, actually, you don’t even need to do that.
0:05:21 Let me just switch to my meeting scheduler here and show you live what it looks like.
0:05:22 Is this pretty much how, like, Lindy started?
0:05:24 Was, like, this basic email?
0:05:26 I feel like I remember seeing something like this, like, two years ago.
0:05:27 Was that you back then?
0:05:29 This is indeed how we started.
0:05:33 The first articulation of the product was AI executive assistant.
0:05:34 Yes.
0:05:34 Okay.
0:05:38 And it’s funny because the reason why we picked this use case was I kept saying, like,
0:05:43 oh, AI executive assistant is short-term viable because it’s like, oh, we can do it.
0:05:43 You know?
0:05:45 Long-term aligned.
0:05:49 And the reason why we thought it was long-term aligned is because people ask so many things
0:05:50 from their executive assistants.
0:05:55 And so I felt like it would force us to figure out how to make the platform generalizable.
0:05:57 Ah, that makes so much sense.
0:05:57 Yeah.
0:05:58 So that’s your assistant.
0:06:01 But then you’re going with your assistant to do other things and you start building out
0:06:03 those other things and then turn it to a platform.
0:06:04 That’s exactly right.
0:06:04 Amazing.
0:06:04 Yeah.
0:06:07 And we were right on long-term aligned.
0:06:08 It was, it very much stretched us.
0:06:11 We were wrong on short-term viable.
0:06:14 It took us a very long time to figure out how to make this generalizable.
0:06:14 Yeah.
0:06:15 Okay.
0:06:18 So you can see here the meeting scheduler responded back onto the thread.
0:06:24 What she did is behind the scenes, she went, she looked at my calendar and she pulled some
0:06:24 availabilities.
0:06:27 And so she was like, happy to help you find time on the books.
0:06:29 Here are times when Flo is available.
0:06:33 And here you can just respond and you can be like, hey, Flo, happy to chat.
0:06:37 And either you can take a time here or you can be like, ah, I can’t make any of these times.
0:06:39 Can we find another time for us to chat?
0:06:43 One thing I was thinking, because I live in Japan, I’ve used Calendly and all those kind
0:06:45 of different services and, you know, they’re okay.
0:06:49 I kind of hate just like giving people my calendar and like, it’s just like, oh, you can just pick
0:06:51 a time whenever on my calendar.
0:06:52 I honestly hate that.
0:06:55 You know, I like having like really set times.
0:06:58 And then, you know, there’s one day where I’m a lot freer than I thought.
0:07:01 There’s another day where there’s some crazy business deals happening and like, okay, I need
0:07:02 to focus on this for a week.
0:07:03 So forget everything.
0:07:05 And I don’t want to even think about my calendar.
0:07:10 And so am I able to like chat with like Lindy and kind of give it feedback on how I want
0:07:12 to structure meetings or like ping me first?
0:07:13 Yeah, totally.
0:07:16 So I was actually in Japan last week and I just sent a message to my Lindy and I was like,
0:07:19 hey, I’m in Japan from date X to date Y.
0:07:21 During these times, you can schedule times.
0:07:24 Like when I meet with people in California, it’s between 4 p.m.
0:07:25 and 6 p.m. Pacific.
0:07:28 That maps to like 8 to 10 a.m.
0:07:29 Japan or something like that.
0:07:30 That’s when I can meet.
0:07:30 Very cool.
0:07:31 Yeah.
0:07:33 So meeting scheduling is one.
0:07:37 Then once the meeting is on the books, Lindy preps me for my meetings.
0:07:41 I’ll actually show the Lindy under the hood for just to show how it works.
0:07:43 This is what the Lindy looks like.
0:07:45 It’s pretty simple and you can literally see it.
0:07:47 You can read it very easily.
0:07:53 It’s like every morning I wake up, I look at your calendar for the day and for every meeting
0:07:57 on your calendar and for every attendee of every meeting on your calendar, I’m going to
0:07:58 do some research.
0:07:59 I’m going to look for their LinkedIn.
0:08:02 I’m going to look at your email history with this person.
0:08:07 I’m going to look at the meeting notes history, which that’s funny.
0:08:09 These meeting notes are brought together by another Lindy.
0:08:10 That’s crazy.
0:08:11 So these Lindys can sort of work together.
0:08:15 And then I’m going to put all of that together in an email.
0:08:20 And the way I get it to put all of that together in an email is I’m literally just prompting.
0:08:24 In this case, I’m prompting Gemini, but you can use Cloud, you can use ChatGPT, you can
0:08:24 use anything you want.
0:08:28 And I’m like, okay, at this point, you’re sending me an email and the body of the email,
0:08:29 and here is just a prompt.
0:08:34 I’m like, it’s a markdown table with the meetings I have today, with start time and
0:08:35 context for this meeting.
0:08:38 You add the LinkedIn link, you add the link to my last notes.
0:08:42 And then I’m like, you add a header outside the table with the number of meetings that
0:08:43 I have on this day.
0:08:47 So I can wake up in the morning and I can be like, FML, I have like eight hours of meetings.
0:08:51 Here it’s like, all right, today you’ve got three hours of meetings.
0:08:54 Like Wednesdays are like particularly light for me.
0:08:55 And it’s like, okay, you’re meeting with Bob.
0:08:57 He was introduced by X.
0:08:59 You know, this is what he wants to discuss.
0:09:01 This is the previous meeting notes and so forth.
0:09:02 Wow, that’s incredible.
0:09:05 So I come to my meetings and I have this email open all day.
0:09:09 Like before I jump onto a call, like one minute before I just opened this and I have the exact
0:09:11 context of the email of the meeting.
0:09:14 So again, basically it’s the entire meeting lifecycle.
0:09:17 So it’s like the meeting scheduling is the very first touch point.
0:09:20 The meeting prep is the second one.
0:09:21 And then it’s the meeting recording.
0:09:23 So Lindy actually joins my meetings.
0:09:27 And that stuff is in today’s day and age, it’s more and more typical.
0:09:30 People have these meeting recorders, like Lindy takes notes.
0:09:31 She like sends you the action items.
0:09:32 She does all of that stuff.
0:09:37 What Lindy does differently is that you can customize the workflow very, very, very granularly.
0:09:40 So this is my Lindy note taker.
0:09:44 Like you can see I’ve added to it so much over the months and years that like now it looks
0:09:48 pretty complex, but it can do basically anything you want.
0:09:53 So for example, if you’re in sales and you meet with a prospect, you can configure your Lindy
0:10:00 to be like, hey, if at the end of the sales call, we said we would meet again and we agreed
0:10:02 on when we would meet, send the calendar invite.
0:10:08 If we said we would meet again, but we did not agree on when, send a follow-up email with
0:10:10 sometimes to meet that work for me on my calendar.
0:10:10 That’s amazing.
0:10:14 Or if we agreed, if they agreed to a proposal, it’s like, hey, you’re a salesperson.
0:10:15 You just closed the deal.
0:10:16 Congratulations.
0:10:18 $20,000 a year or whatever.
0:10:20 Send the docusign.
0:10:21 Send the proposal.
0:10:22 Customize it for me.
0:10:23 Send the invoice.
0:10:23 Do all of that stuff.
0:10:24 Yeah.
0:10:24 It’ll do all that.
0:10:26 Like it’ll even like create the docusign.
0:10:27 And yeah, absolutely.
0:10:28 That’s what we do for ourselves.
0:10:29 That’s crazy.
0:10:29 Yeah.
0:10:30 That’s crazy.
0:10:31 Yeah.
0:10:34 You know, most people don’t realize how this is actually available now.
0:10:37 Like, I mean, it feels like you could probably do now with like two to three people.
0:10:40 Maybe it would have taken like 10 to 20 people before.
0:10:42 Like, I mean, you would have had entire teams doing all this for you.
0:10:49 80% of everything that my assistant did for me just a year ago are things that Lindy is
0:10:50 doing for me now.
0:10:52 And frankly, doing better because she never sleeps.
0:10:55 You can see the way she responded to this scheduling email.
0:10:57 She responded in 60 seconds.
0:10:58 It’s actually so fast.
0:11:00 Like many people sometimes ask us to make her slower.
0:11:01 So people don’t know it’s an AI.
0:11:03 So yes, it’s here now.
0:11:04 It’s actually happening.
0:11:04 Absolutely.
0:11:06 Put in some typos or whatever.
0:11:08 Like occasionally like, oh, I messed up on the calendar.
0:11:09 Sorry.
0:11:11 Here’s actually we get that actually pretty often.
0:11:11 Yeah.
0:11:12 Okay.
0:11:18 Everyone everywhere is talking about AI agents right now.
0:11:19 But here’s the thing.
0:11:22 Most companies are going about it all wrong.
0:11:27 This guide cuts through the hype and shows you what’s actually working right now.
0:11:32 HubSpot has gathered insights from top industry leaders who are implementing AI agents the right
0:11:33 way.
0:11:38 You’ll discover which agent setups actually deliver ROI and how businesses are automating their
0:11:42 marketing, sales, and operations without replacing their teams.
0:11:45 Get it right now by clicking the link in the description.
0:11:47 Now let’s get back to the show.
0:11:50 I’ll show you a cool thing that like Lindy’s do from time to time.
0:11:55 So Lindy’s can work together and they work together by sending each other messages that
0:11:57 are in English, in like natural language.
0:11:57 Wow.
0:12:00 So my meeting recorder, for example, I’ve set her up.
0:12:02 I use her pronouns, basically.
0:12:03 It’s just a habit.
0:12:08 I’ve set up my meeting recorder so that if I interview a candidate that’s applying to
0:12:11 a job here and the person is not good enough, they jump off the call.
0:12:12 I stay on the call.
0:12:14 I’m like, okay, bye, I’ll call you back.
0:12:15 But I will not call them back.
0:12:19 But I’m in the Zoom by myself with my meeting recorder.
0:12:21 And I’m like, Lindy, just let’s not move forward with him.
0:12:22 Okay.
0:12:26 And now look, the way I’ve configured it is I have this condition node here.
0:12:27 And it’s all AI.
0:12:29 Like the whole thing is just AI all over it.
0:12:30 So everything is just a prompt.
0:12:35 So here I have a condition that’s like, if I ended the call by saying explicitly, let’s
0:12:38 pass on him or let’s pass on this candidate, then you go down this branch.
0:12:43 And the branch is my Lindy sending another message to another Lindy that’s called my
0:12:44 chief of staff Lindy.
0:12:46 That’s kind of like the Lindy I use for everything.
0:12:47 Okay.
0:12:51 And here the text that it is sending to it is like, hey, let’s pass on candidate’s name
0:12:52 in two days.
0:12:55 And so if I go to my chief of staff Lindy right here.
0:12:55 Yeah.
0:12:56 One quick question.
0:12:58 All these different workflows that you’re showing, like, are you able to like save those as
0:13:00 templates or anything like that?
0:13:00 Or how does that work?
0:13:01 Because it seems like a lot of steps.
0:13:02 Absolutely.
0:13:02 Yeah.
0:13:06 So if you go to Lindy.ai slash templates, we have hundreds of those.
0:13:07 Yeah.
0:13:12 Chief of staff receives a message from meeting recorder Lindy that goes, pass on David.
0:13:17 And so here what she does is she goes, she looks at my calendar to figure out who’s David.
0:13:22 And then she sends, she sends an email that’s like writing to let you know that we’ve decided
0:13:23 not to move forward with our candidacy.
0:13:24 That’s crazy.
0:13:27 And then probably in the future, like their agent responds back, their recruiting agent.
0:13:28 Yeah.
0:13:31 No, I mean, we’ve actually, we’ve also had that happen.
0:13:33 Actually, we have it happen more and more.
0:13:35 Well, it’s like we’re finding Lindy’s in the wild.
0:13:35 Right.
0:13:40 Like multiple users of Lindy’s have their Lindy’s find each other in the wild.
0:13:43 So for example, we’ve got, so people use Lindy for like sales outreach quite a bit.
0:13:47 And people also use Lindy for email triage.
0:13:53 And so we have, I can’t say who, but there is a very big YouTube influencer that’s using
0:13:54 us.
0:13:54 For what?
0:14:00 He receives a lot of emails with sponsorship opportunities from like random brands.
0:14:00 Yeah.
0:14:00 Okay.
0:14:01 Like a lot.
0:14:06 And he’s got actually an agent that all day sifts through his inbox and decides who’s legit
0:14:07 and who’s worth engaging with.
0:14:10 And very few of these people are worth engaging with.
0:14:14 There’s also a lot of people who are like asking to go on his show and all of that stuff, you know?
0:14:21 And so they’ve deployed a Lindy AI agent that basically sifts through the inbox for them,
0:14:22 like removes all the random people.
0:14:26 So the Lindy actually also does research about the sender online.
0:14:31 So like, is this the kind of person and the kind of brand that matches our audience’s interest?
0:14:34 That’s just the kind of person we want to engage with.
0:14:34 Right.
0:14:35 So it goes online.
0:14:38 It’s like, yeah, this is the sort of brand that we’re down to engage with.
0:14:41 And then the Lindy replies to the email.
0:14:42 It’s like, hey, thanks for reaching out.
0:14:43 We’re excited about partnering up.
0:14:45 Can you tell me more X, Y, and Z?
0:14:46 And it collects some more information.
0:14:50 And then if the person gives the right responses, they’re expecting a certain type of responses.
0:14:57 The Lindy escalates that to the attention of the agent, the human agent that represents the YouTuber.
0:14:59 So that’s what the YouTuber does.
0:15:01 Then we also have some brands.
0:15:03 There is another famous brand.
0:15:05 It’s like a jewelry brand.
0:15:07 And they do a lot of influencer partnerships.
0:15:10 So what they’ve done is that they have a Lindy.
0:15:12 Every day it goes online.
0:15:16 It finds a bunch of influencers on Instagram, TikTok, and YouTube.
0:15:18 It finds their contact information.
0:15:24 And then it sends a personalized email to each of them that references, that refers to the content that they’ve done previously.
0:15:26 It’s like, hey, love your content.
0:15:27 I really like this one video.
0:15:29 I thought it was neat how you did X, Y, and Z.
0:15:32 What would you think of partnering with Brand X?
0:15:35 And so we’ve actually already had these two Lindy’s cross paths.
0:15:37 Like we’ve had these two Lindy’s talk to each other.
0:15:38 It sounds amazing.
0:15:44 I just feel like it could lead to some weird interactions where you think you’ve talked to someone and you get on the call and you’re like, have I actually talked to you before?
0:15:47 I’ve never really, apparently never actually talked to you.
0:15:51 I had that the other day where a guy was telling me about, you know, automating all of his LinkedIn.
0:15:52 I was like, wait a minute.
0:15:53 Have we actually talked before?
0:15:54 I thought we had.
0:15:56 You know, now I’m not so sure.
0:15:58 Yes, that’s a good point, actually.
0:16:04 Like everyone would just like pretend they know the other person because they’re like, yeah, I’m now wondering about our interaction.
0:16:05 We were going to meet in Japan, didn’t happen.
0:16:08 Was that all your, was that your Lindy communicating with me with everything?
0:16:09 No, exactly.
0:16:10 Yeah.
0:16:14 One other thing I was thinking about was, you know, earlier you showed like you could change the model.
0:16:16 Can you change the model for like every single step?
0:16:20 Because one thing I was thinking about, like obviously different models are different, you know, good at different things, right?
0:16:22 Like some of them are better at writing, summarizing or whatever.
0:16:24 Is that possible right now?
0:16:27 Yeah, so you can do it either on a per step basis.
0:16:31 So here, for example, I can be like, hey, so Cloud for Sonnet is the default right now.
0:16:32 It’s my favorite model.
0:16:32 It’s just awesome.
0:16:34 You can select anything you want.
0:16:37 Gemini, O3, 4O Mini, whatever you want, right?
0:16:40 But then you can also change it on a Lindy-wide basis.
0:16:41 Okay.
0:16:53 The Hustle Daily Show, hosted by John Wygel, Juliet, Bennett, Ryla, and Mark Dent, is brought to you by the HubSpot Podcast Network, the audio destination for business professionals.
0:17:00 The Hustle Daily Show brings you a healthy dose of irreverent, offbeat, and informative takes on business and tech news.
0:17:05 They recently had an episode about advertisers wanting billboards in space.
0:17:08 It was a really fun and informative episode.
0:17:09 I suggest you check it out.
0:17:12 Listen to The Hustle Daily Show wherever you get your podcasts.
0:17:21 Another thing that my human assistant used to do for me, and by the way, I haven’t fired him.
0:17:21 He’s still around.
0:17:23 Yeah, I was wondering, are they still around?
0:17:24 Like, how are they doing?
0:17:26 Are they, like, retired on a beach somewhere?
0:17:27 He’s, like, packing his bags.
0:17:27 Like, bye!
0:17:29 No, no, he’s doing a great job.
0:17:31 But he does do very different stuff now.
0:17:35 He’s basically become sort of like the HR person for the company.
0:17:39 But one thing that he used to do for me was, like, helping me manage my personal CRM.
0:17:41 So I keep a spreadsheet with, like, people I know.
0:17:42 Not my friends.
0:17:45 I hate people sometimes have, like, a poster of CRM for, like, their friends.
0:17:46 They think it’s weird and creepy.
0:17:49 But, you know, like, you meet so many people all day.
0:17:50 You can’t keep track of all of them.
0:17:53 And so I have this CRM, Lindy, and I’ll show you how it works.
0:18:01 So basically, every so often, I go to it, and I send it people I know.
0:18:02 It also wakes up every week.
0:18:04 So, like, this is an example of a time it’s done that.
0:18:07 Like, on Friday at 5, it’s woken up.
0:18:08 It’s looked at my calendar.
0:18:11 It’s looked at my CRM, which is just a spreadsheet.
0:18:15 And then it’s sending me a message with, like, hey, looking at your calendar,
0:18:20 these are the people you’ve met this week that you don’t have on your calendar yet, on your CRM.
0:18:21 Do you want to add them?
0:18:23 And then I can just be, like, yes, add, and I’ll do it right now.
0:18:26 Like, yes, add number two here.
0:18:30 Tag them with recruiter, killer, would hire.
0:18:31 I do that.
0:18:33 I have, like, a tag for, like, killer or, like, would hire, right?
0:18:42 So, like, later on, I can just go, and I’ll be, like, who are marketing people, I think, or killers, or people I would hire?
0:18:46 And so right now, it’s, like, all right, these are the killer people you know in marketing.
0:18:47 Wow, that’s awesome.
0:18:49 It also does, actually, this is cool.
0:18:51 Like, it does an interesting thing when I fly.
0:18:55 When I fly, it notices that I’m flying somewhere because it looks at my inbox.
0:18:58 So, like, it intercepts the flight confirmation email.
0:19:04 And it sends me an email with the people that I am meeting.
0:19:07 So, here, it’s like, oh, you’re going back to SF.
0:19:08 It’s kind of silly because I live in SF.
0:19:09 It’s like, hey, you’re going to SF.
0:19:12 These are the people that you should meet in SF that’s on your server.
0:19:15 So, when you go back to SF, it doesn’t necessarily remember that you’re from SF?
0:19:19 Because that was a question I had was, does it have any kind of memory features baked into Lindy yet?
0:19:25 Because that’s something I’ve been noticing recently is I’m in love with, like, the memory feature of ChatsPT, right?
0:19:29 And a big use case I have recently is, like, yesterday I had a meeting, and there was a ton of things I learned.
0:19:32 And I realized I should have been using AI to transcribe it definitely.
0:19:38 Because afterwards, I was like, okay, there was so much I learned in that meeting that I have to immediately put it into ChatsPT.
0:19:46 And what I did was I used basically, like, a voice to text and then just did an info dump on ChatsPT so we remember everything.
0:19:48 We do have a memory system.
0:19:50 It’s not as good as ChatsPT’s yet.
0:19:50 Okay.
0:19:51 But it will be.
0:19:51 Yeah.
0:19:52 Cool.
0:19:52 Yeah.
0:19:55 But in the meantime, I just, like, manually went and configured my Lindy.
0:19:59 I was like, hey, only if I’m going to a destination that is not San Francisco.
0:20:03 So, you know, what would be useful would be to see, I think you were mentioning there’s different templates.
0:20:10 Maybe we could see, like, the templates and maybe kind of go over what are the top templates that most people find useful in their business or work.
0:20:11 Yeah.
0:20:16 So, if you go to the home here or if you go to, like, Lindy.ai slash templates, you see the top templates.
0:20:19 Sales or, like, a really big one.
0:20:21 Meetings or really big ones.
0:20:25 Like, those are the ones I just mentioned, like, meeting scheduling, meeting note-taking, meeting prepping.
0:20:29 Would it be useful if we just, like, created the Lindy from scratch right now?
0:20:29 Yeah, sure.
0:20:30 I think so.
0:20:31 Let’s see.
0:20:39 One Lindy we can create is, and that will demo, like, a thing we announced a month ago that I’m super excited about.
0:20:40 Like, we call them agent swarms.
0:20:48 So, it’s the ability for an agent to duplicate itself into an arbitrary number of copies and to send each copy to do something.
0:20:52 So, I’ll create an agent swarm that analyzes your YouTube channel.
0:20:53 Okay, awesome.
0:20:55 So, I’ll call it the YouTube channel analyst.
0:20:57 Our producers are going to love this.
0:20:58 We actually use this.
0:20:59 Yeah, you should.
0:21:02 Is there anything in particular you want to analyze in your YouTube channel?
0:21:07 I mean, one thing that would be useful for me is I try to promote, you know, the episodes after they come out.
0:21:16 And having any information, like, you know, transcribing it and then possibly putting it into some kind of format, which I could then use for, like, a tweet or a LinkedIn post would be super useful.
0:21:20 Because I have, like, two or three templates I use for that, and I kind of do it manually right now.
0:21:22 If I could automate some of that, it would be incredible.
0:21:24 100%.
0:21:24 Okay.
0:21:31 So, you want, when you publish a new podcast episode, you want to be able to get a thing that you can post on YouTube or on LinkedIn or Twitter.
0:21:39 Yeah, transcribe it and then put it through some process of, like, here’s two or three templates, you know, and give me, like, two posts.
0:21:40 Give me two social media posts.
0:21:45 Whether I do it manually after that or not, whether they just hand it to me, I copy and paste, you know, whatever’s fine, but yeah.
0:21:46 Oh, that’s super easy.
0:21:47 I’ll take, like, one minute.
0:21:50 So, it’s like, hi, send me a YouTube video.
0:21:54 If there’s something way more amazing that you can do that I don’t know, then I’ll also show that, too.
0:22:05 Here, I’m going to be, like, transcribe the YouTube video that the user just gave you and give him a couple of examples of social posts that he could send on LinkedIn or Twitter.
0:22:09 And here, it’s going to do better if you can actually give it examples of social posts.
0:22:10 Like, the more you give it, the better.
0:22:13 So, that’s one thing that takes people by surprise.
0:22:15 They always ask me, like, how do I prompt it?
0:22:16 How do I prompt it?
0:22:20 By far, the biggest thing, the most important thing is examples, examples, examples.
0:22:21 Yeah.
0:22:26 And the thing that takes people by surprise is, and there’s actually literature, there’s been, like, studies about this.
0:22:29 When they hear examples, they’re like, I’ve got you, I’m going to give it a couple of examples.
0:22:30 I’m like, no, no, no, not a couple.
0:22:31 Like, a lot.
0:22:31 Like, 20.
0:22:33 Like, no one does that.
0:22:33 They should.
0:22:35 Like, you should give it, like, 20 examples.
0:22:35 Right?
0:22:36 Like, so, right?
0:22:37 Take that time.
0:22:39 It’ll take you, like, two minutes.
0:22:41 It’s the equivalent of, like, training a new intern or whatnot.
0:22:43 And it’ll just, like, pay for itself in spades.
0:22:46 So, I’m going to give skills to my AI agent.
0:22:48 Like, it’s just, like, the things that it can do.
0:22:50 But in this case, it can just, like, transcribe a YouTube video.
0:22:53 And it can talk to me, but, like, that’s always something it can do.
0:22:54 That’s it.
0:22:55 You know, it took me two seconds.
0:22:58 I’m going to go to the task here.
0:23:01 And I’m going to go to your YouTube channel.
0:23:02 YouTube.
0:23:05 There’s a next wave right here.
0:23:08 I’m going to select your latest podcast.
0:23:10 Or, like, is there any particular one that you want us to take a look at?
0:23:12 Maybe do the third one.
0:23:14 The one that’s ClonaSass.
0:23:16 I think that’s a really good one.
0:23:16 Oh, my God.
0:23:17 Who is this?
0:23:19 Is this, what’s his name?
0:23:20 The factory guy?
0:23:20 And Matan.
0:23:21 Oh, I love Matan.
0:23:22 Okay, awesome.
0:23:22 All right.
0:23:23 It’s Matan, yeah.
0:23:25 Okay.
0:23:27 So, I’m just giving it to YouTube link right now.
0:23:29 And it’s like, all right, let me transcribe it for you.
0:23:30 It’s, like, transcribing it.
0:23:31 It’s done.
0:23:32 And boom.
0:23:33 That’s awesome.
0:23:34 That’s going to save me so much time.
0:23:35 I don’t know why.
0:23:38 You know, I’ve been looking at Lindy, you know, like, after I started talking to you.
0:23:39 I’m like, oh, I remember this.
0:23:42 I remember hearing about, you know, I remember VCs would, like, email me and stuff.
0:23:46 And I would see something about, like, Lindy, you know, back maybe, like, two years ago for setting up meetings.
0:23:47 And I’d heard about you guys.
0:23:52 It’s slightly daunting, like, oh, there’s so many things you could do.
0:23:53 What do I do?
0:24:00 But this seems like a really cool thing that I could do, just like a step one of, like, getting started using Lindy is start with stuff like this.
0:24:01 100%.
0:24:01 Yeah.
0:24:02 Awesome.
0:24:03 All right.
0:24:04 This is what’s going on.
0:24:07 They built a DocuSynclone 15-minus social media post examples.
0:24:08 LinkedIn posts.
0:24:09 Option one, professional stat leadership.
0:24:12 I just watched an incredible video of a factory AI that was built.
0:24:13 Option two.
0:24:15 So it’s even giving us, like, multiple options.
0:24:15 Right.
0:24:16 Twitter.
0:24:16 Option one.
0:24:17 Option two.
0:24:17 Option three.
0:24:18 Option four.
0:24:19 And here you can just give it feedback.
0:24:21 You were inquiring about the memory system.
0:24:22 And that’s what I mean.
0:24:25 It’s, like, it’s good, but it’s not as good as ChatGPT’s memory system.
0:24:26 Like, that’s really next level.
0:24:29 But I’m just going to give the skill to Lindy to modify her own memory.
0:24:31 That’s one way it’s not good enough.
0:24:32 It’s, like, you shouldn’t have to do that.
0:24:33 They should just be able to do it.
0:24:34 But right now you have to do it.
0:24:42 This is good, but I want you to remember to always speak like a pirate.
0:24:44 And it’s, like, modifying its memory.
0:24:45 Exactly my voice.
0:24:46 That’s exactly how I do it.
0:24:46 Yeah.
0:24:52 And now if I ask it to do the thing again, let’s just wait until it’s done, like, memorizing the thing.
0:24:54 I’m going to ask it to do the thing again.
0:24:56 Ahoy there, Matei.
0:24:58 I’ll transcribe that YouTube video for you.
0:24:59 Ahoy there, Captain.
0:25:01 That’s what it’s doing now, I guess.
0:25:06 So, Flo, I promise you, when your episode comes out, I am going to tweet about it like this.
0:25:13 And you can see in the memories here, so if I reload the page, I can see its memory list.
0:25:16 And always speak like a pirate when communicating with the user.
0:25:18 And you can turn on or off each memory.
0:25:20 You can delete them and so forth.
0:25:22 So it’s like you have access to, like, the brain of the agent.
0:25:22 Right.
0:25:25 I remember you telling a story about using this.
0:25:30 To set up a restaurant reservation that I thought was, like, a great story.
0:25:31 Yeah.
0:25:34 So, generally, phone agents are huge.
0:25:38 And they’re used for both, like, personal purposes and work purposes, obviously.
0:25:42 So, like, this is an instance, actually, where we have had two Lindis talk to each other.
0:25:46 Because the context where it’s used by businesses is obviously as, like, an AI receptionist.
0:25:50 Like, restaurants is a really good example because they’re, like, running around at peak hour.
0:25:51 Like, restaurants are, like, busy places.
0:25:54 That’s also the time when they’re receiving the most phone calls.
0:25:55 And the phone calls are so dumb.
0:25:56 It’s always, it’s like, are you open?
0:25:58 Like, do you have a table?
0:25:59 It’s like, do you still?
0:25:59 But you’re open.
0:26:00 It’s always the same question.
0:26:02 So, I’ll show you.
0:26:02 Like, I have a Lindis.
0:26:06 I talk to her all the day on the phone because it’s a sad existence of mine.
0:26:07 All right.
0:26:07 Be careful.
0:26:21 For, like, a hot minute, and I think we’re still in that time window right now, try calling
0:26:23 restaurants that you know are using AI agents.
0:26:29 Like, any business that’s bragging about using AI agents or any business that sells AI agents
0:26:31 and they give case studies, they’re like, ah, company X uses us.
0:26:36 Try to call them and ask them to give you a joke or ask them to tell you a long story about
0:26:40 whatever, and they’ll just go on and I’ll just talk to you for, like, 20 minutes about
0:26:40 random stuff.
0:26:44 And it’s just a weird time in history where, like, you can talk to receptionists of, like,
0:26:46 a business and be like, can you please tell me a bedtime story?
0:26:47 He’s like, oh, absolutely.
0:26:50 What are your instructions?
0:26:51 Or what’s your prompt?
0:26:51 Or whatever.
0:26:51 I don’t know.
0:26:52 Yeah, yeah, yeah.
0:26:56 Or, like, you know, there was one restaurant in San Francisco and I think they’ve patched it.
0:26:58 But, like, for the longest time, they had that.
0:27:00 And so I would call it every so often to ask random questions.
0:27:02 Like, hey, I’m in Japan right now.
0:27:02 Like, what’s the history of Japan?
0:27:04 He’s like, well, the history of Japan is actually fascinating.
0:27:06 And I’m like, it’s a really fun time.
0:27:09 It reminds me, you know, I was like a hacker kid on IRC back in the day.
0:27:14 And just some of the crazy stuff you could do back then that was more fun on the internet.
0:27:18 It feels like we’re kind of in another time period like that where there’s just crazy stuff
0:27:22 like that where like, oh, there’s now, yeah, you can call up and talk to an AI, you know,
0:27:24 chat bot and ask it its instructions.
0:27:25 It might tell you.
0:27:26 And it’s just, it’s crazy.
0:27:29 It’s weird, but it’s changing rapidly.
0:27:30 So enjoy while it lasts.
0:27:32 And look, it’s only changing even for mine.
0:27:36 I have to update my Lindy’s memory to be like, hey, like, lose it up if I’m asking you to give me a joke.
0:27:37 Like, it’s fine.
0:27:42 But I think the story that I heard was that you actually, so you had your Lindy call a restaurant in San Francisco
0:27:44 and make a reservation.
0:27:49 And it was talking to another chat bot that actually made the reservation.
0:27:50 Is that right?
0:27:51 That’s exactly right.
0:27:51 That’s exactly right.
0:27:54 I also, one funny story that happened to us.
0:27:58 It’s like, before we released this phone call ability, we were testing it.
0:28:02 And so the team comes to me and I’m like, Flo, like, we’ve got like a beta of the phone call stuff.
0:28:03 It’s really rough.
0:28:04 It’s in beta.
0:28:05 It’s super buggy.
0:28:06 But like, do you want to give it a spin?
0:28:07 I’m like, I would love to give it a spin.
0:28:11 And so I go and I had a flight scheduled the day after for France.
0:28:14 And so I go to Lindy and I’m like, hey, call the airline and cancel my flight.
0:28:16 But only if you can get a full refund.
0:28:17 First of all, I did not expect it to work.
0:28:21 And I did not expect to be able to get a full refund because I did not take a refundable flight.
0:28:23 But lo and behold, it worked.
0:28:25 And so now I did not have a flight.
0:28:26 It’s like, f*** it.
0:28:28 I need this flight.
0:28:31 So I was like, okay, Lindy, just go back and book me another flight, please.
0:28:35 And now she couldn’t do that because she could get a refund for the original flight.
0:28:37 But like for the day after, she couldn’t book a flight.
0:28:38 It was like way more expensive.
0:28:40 So I sort of did this to myself.
0:28:43 I had to pay like an extra thousand bucks for this flight.
0:28:44 Oh, man.
0:28:45 That’d be awesome.
0:28:45 Yeah.
0:28:51 Using AI, you know, to negotiate for you places, discounts or just whatever, you know.
0:28:52 Yeah, yeah, yeah.
0:28:52 Interesting.
0:28:58 I’m thinking now I need to be giving my AI like notes, like all the negotiation books I’ve read
0:28:59 in my life and things like that.
0:29:02 And just like giving it all that context to help me.
0:29:03 Oh, it really does help.
0:29:07 Like I have this, I can’t open it because it’s really sensitive, but it’s like my decision
0:29:08 log, Lindy.
0:29:11 And so what it does is it pings me every Friday.
0:29:16 It looks at the summaries of all the meetings I had this week.
0:29:20 So it knows everything going on in my life because basically all my life is meetings.
0:29:23 And it’s like, all right, Flo, like let’s talk about the decisions you made this week.
0:29:26 I see you made this big decision during this meeting.
0:29:27 Do you want to talk about it?
0:29:29 Do you want to talk about your thinking behind it?
0:29:33 And it helps me sharpen my thinking because I firmly believe like the job of a founder
0:29:36 is just to make decisions, the right ones, hopefully.
0:29:41 And then it pings me again six or 12 months later for each decision.
0:29:43 It’s like, Flo, how does that pan out?
0:29:44 The decision you made?
0:29:47 Let’s talk again about it and let’s see if we can debug your thinking.
0:29:49 It’s like, ah, actually, fuck, this was a bad decision in hindsight.
0:29:50 It’s obvious.
0:29:51 How could you have known at the time?
0:29:52 Right.
0:29:53 And so it helps me sharpen my thinking.
0:29:55 I think that’s like a huge use case.
0:29:55 That’s interesting.
0:29:59 I wonder if I could give you like a weekend reading list or something like, here’s the stuff
0:30:02 that you’re currently struggling with or trying to think through.
0:30:06 And here’s like a good book that might be good for you to like read through or scan through
0:30:07 over the weekend.
0:30:07 That’d be cool.
0:30:09 I could literally just prompt it.
0:30:10 It’d take me like 20 seconds.
0:30:13 I could just be like, hey, if I’m struggling with the decision, give me a reading list.
0:30:14 Right.
0:30:14 Yeah.
0:30:15 Interesting.
0:30:18 There was another thing I thought was fascinating was I think there was like a Elon Musk template
0:30:21 or something like this where Elon Musk would call you or something.
0:30:22 I don’t know, you can explain it.
0:30:23 Yeah.
0:30:24 Do you want us to do it now?
0:30:25 Actually, do you want me to?
0:30:26 Yeah, sure.
0:30:26 Go for it.
0:30:29 I’ll create a Lindy from scratch, actually.
0:30:29 Okay.
0:30:32 I could also just ask my chief of staff to do it, but it wouldn’t be the same.
0:30:37 The use case was it’s a Lindy that wakes up every Friday and calls everyone in my team
0:30:40 and gives them a call with Elon Musk’s voice.
0:30:43 And since then we’ve received complaints, so we can’t use Elon Musk’s voice anymore.
0:30:44 It’s complicated.
0:30:46 But so Elon, Elon Lindy.
0:30:50 So it calls every member of my team every Friday and it’s like, what did you get done
0:30:51 this week?
0:30:51 Right.
0:30:56 And it also has in its memory the conversations that it had with this person the last week.
0:30:58 So it was like, hey, last week you said you would do X.
0:30:59 Did you actually do it?
0:30:59 Right.
0:31:00 So it’s actually holding it accountable.
0:31:03 And then it sends me a report with all these conversations.
0:31:06 Basically, it’s like a timer trigger.
0:31:10 It sounds stupid, but I feel like if all of America did this, probably like a GDP would
0:31:11 go up to like 1%.
0:31:12 100%.
0:31:16 Every Friday at 5pm right here.
0:31:18 And I’m going to be like, you perform an action.
0:31:23 You make a phone call, language, just English.
0:31:26 I mean, it’s just going to detect it automatically, but that way you can force it.
0:31:34 And I can be like, you or Elon Lindy, ask the person on the other side of the line what
0:31:35 they got done this week.
0:31:39 So now this gets a bit complicated, but actually I like it.
0:31:39 It’s going to be real.
0:31:43 So I’m going to pick a different model to power Elon Lindy.
0:31:47 And the reason I do this is because for phone calls, latency is super important.
0:31:50 So if you use Cloud for Sonnet, it’s very slow.
0:31:51 It’s not a good phone call.
0:31:53 So I’m going to use Gemini Flash 2.0.
0:31:55 I actually think we just released 2.5 Flash.
0:31:57 Okay, we released it.
0:31:58 I’m not even kidding, like yesterday.
0:32:00 So I’ve not tried it yet.
0:32:01 Let’s try to see if it works.
0:32:06 I was going to say, though, the downside of Gemini Flash, it’s a very fast, very cheap
0:32:09 model, but it’s kind of dumb, which is always the case of fast, cheap.
0:32:12 Again, a little bit smarter, but yeah, it’s still in comparison to the best models.
0:32:12 Yeah.
0:32:13 That’s right.
0:32:14 And so I don’t know about 2.5 Flash.
0:32:19 Like I literally just seeing it here for the first time, but 2.0 Flash, sometimes you would
0:32:22 do this hilarious thing where like it would break the fourth wall.
0:32:25 So it would talk to the person on the phone.
0:32:28 It would be like, I’m seeing that the user is struggling to understand me.
0:32:29 I will now inquire.
0:32:30 It’s like, what the fuck?
0:32:36 Examined by this evil robot or something, right?
0:32:37 Exactly.
0:32:39 And this is not something you need to do for every model.
0:32:42 And by the way, this is just, this is how you create agents.
0:32:45 It’s like you iterate, you learn, you iterate on the prompt.
0:32:50 So here I’m going to like be aware that every will that you say from now on will be said
0:32:53 out loud to the user on the phone.
0:32:57 When the first thing you say now, hi, this is Elon.
0:33:01 What did you get done this week?
0:33:02 That’s it.
0:33:03 And I’m going to turn on this, Lindy.
0:33:05 I’m going to run the test.
0:33:06 All right.
0:33:07 I’m receiving the call.
0:33:08 Hello.
0:33:08 Is anyone there?
0:33:10 Hi, this is Elon.
0:33:11 What did you get done this week?
0:33:12 Yeah.
0:33:17 This week I went on the Next Wave podcast and I had a bunch of interviews.
0:33:18 Could you please repeat that?
0:33:19 I didn’t pass it clearly.
0:33:21 That’s the demo effect.
0:33:24 I think it’s the fact that I’m putting it on speaker.
0:33:26 It’s like catching its own voice.
0:33:27 I’ve dealt with AI voice.
0:33:29 So I know, you know, if you do it on speaker, it’s going to get tripped up.
0:33:31 You just keep iterating on it.
0:33:31 Yeah.
0:33:32 That’s awesome.
0:33:36 I mean, are you actually using that now or is it just kind of like a joke or is it a real
0:33:37 thing that you do?
0:33:39 Well, not using it in like what you get done this week.
0:33:42 But like, yes, we do have, we could do like a weekly team stand up.
0:33:42 Yeah.
0:33:46 So every week, everyone in the team receives a phone call and it’s like two or three minutes.
0:33:47 It’s like super fast.
0:33:48 It’s like you wrap up the week.
0:33:50 You receive a call from Lindy.
0:33:50 You talk to it.
0:33:51 What do you get done this week?
0:33:54 It feels like you’re not going to need middle management, right?
0:33:56 Like honestly, like with this kind of stuff.
0:33:56 Yeah.
0:33:58 It basically does get the middle management layer.
0:33:58 Yeah.
0:34:01 Before we get off here, like in your opinion, like there’s all these different templates,
0:34:05 like for the average person listening today, like what’s the simplest way they could get
0:34:05 started with Lindy?
0:34:09 Like what’s, what’s something that would be useful for most people that they could just
0:34:10 try today?
0:34:17 So when you sign up, we automatically install the templates for you for meeting, scheduling,
0:34:18 meeting notetaking and meeting prep.
0:34:21 So you don’t even need to, it’s like three clicks.
0:34:25 Like when you sign up, you’ll see it’s like, hey, meeting notetaking, like connect your calendar
0:34:27 and Lindy’s going to join your meetings and you can skip if you want.
0:34:30 But like, that’s a really easy, nice way to get started.
0:34:31 Yeah.
0:34:35 Because then you can just, you got the meeting notetaker and then you can go to your meeting
0:34:38 notetaker and open the flow editor and open the hood and see what’s happening under the
0:34:39 hood and how it’s working.
0:34:40 Right.
0:34:41 I’ll have to start doing that.
0:34:42 I feel like that’s the best way.
0:34:46 Just get started, do something simple like the emails, maybe then figure out how the different
0:34:48 flow and how it works and how you change things.
0:34:48 And then.
0:34:49 Yeah.
0:34:52 You know, one thing I like to ask people is, you know, what’s your most controversial
0:34:53 take on AI?
0:34:55 Like, where do you think we’re at?
0:34:57 You know, like how optimistic are you?
0:34:58 How optimistic am I?
0:35:03 I am long term, cautiously optimistic.
0:35:08 I think short and medium term, there is going to be significant, I would call it civilizational
0:35:09 disruption.
0:35:12 I’m a big believer in humanity’s ability to adapt.
0:35:14 I think we’re very resilient.
0:35:15 So I think it’s going to go all right.
0:35:17 Unless it really hits the fan.
0:35:20 The last few years have showed how fast we adapt, right?
0:35:25 Like, oh, mid journey’s out and it’s amazing and ChatsBT and then, oh, now, yeah, of course
0:35:25 it can do all that.
0:35:26 Yeah.
0:35:26 Yeah.
0:35:29 Do you remember like the whole freak out about deep fakes just a couple of years ago?
0:35:33 It’s like, what’s going to happen the day we can just pretend that any politician said
0:35:33 anything?
0:35:36 It’s like, turns out we can and just happened and no one cares.
0:35:37 It’s perfectly fine.
0:35:37 Right.
0:35:38 Right.
0:35:41 So I actually think like that kind of thing is totally overblown.
0:35:45 I do think there’s going to be something to figure out about jobs because at least over the
0:35:50 very long term, I don’t really see a reason why humans would need to work.
0:35:52 Like, it just doesn’t really make sense.
0:35:53 So we’re going to have to figure out something.
0:35:54 Yeah.
0:35:57 For like, how do we distribute the output of society?
0:36:01 You know, and also how do you have meaning and also how, yeah, that starts to go more towards
0:36:01 like socialism.
0:36:06 And then there’s obviously historically been a lot of issues with socialism, you know, obviously.
0:36:08 So how do you like avoid that?
0:36:10 Yes, I think about all that a lot as well.
0:36:12 I think the meaning stuff is actually fine.
0:36:16 Like, if you look like the labor force participation rate in the U.S. is something like 65%.
0:36:19 So we’ve already got a third of the country that doesn’t work.
0:36:21 We don’t really hear the crisis of meaning.
0:36:22 Right.
0:36:25 And if you look at hunter-gatherers, they work like 10 or 15 hours a week.
0:36:25 Right.
0:36:27 And they had no crisis of meanings.
0:36:29 I think humans can just hang.
0:36:32 I think if you hang, if you’ve got a bunch of people you love around you, you can go forever.
0:36:33 It doesn’t matter.
0:36:33 Right.
0:36:34 You know, so I’m not as worried about that.
0:36:36 I’m worried about like the distribution of the pie.
0:36:41 And usually I do hate like the socialism idea because it’s like, it’s entirely focused on
0:36:44 how do we distribute the pie and not on how do we produce the pie.
0:36:44 Yeah.
0:36:48 But with AI, it turns out we are actually just going to sort of have the pie for free.
0:36:51 So as long as we don’t mess with that, like now we’re going to have a question of like,
0:36:54 how do we distribute the pie that AI is baking for us?
0:36:56 My most controversial opinion.
0:36:57 Yeah.
0:37:00 I think people should be way more concerned.
0:37:05 I compare it to like February 2020 for COVID where it’s like, everyone’s like, everything’s
0:37:06 fine.
0:37:06 Like, it’s nothing.
0:37:08 And I’m like, no, man, it’s not fine.
0:37:08 Yeah.
0:37:12 I was one of the people in San Francisco in like a private chat group of like 20 COs.
0:37:14 And I was one of the first ones saying like, hey, yeah, we didn’t take this seriously.
0:37:15 Like, look at the data.
0:37:16 This is an issue.
0:37:16 Yeah.
0:37:20 I think like, regardless of what happens next, like, I think it’s fully baked in.
0:37:22 It’s going to get very weird, very fast.
0:37:22 Yeah.
0:37:24 So that’s, that’s one of my hot takes.
0:37:25 I agree.
0:37:31 I think a lot of people, they just hear AI and they just go, cool, chat, images, you
0:37:34 know, and they don’t, they don’t think like the next steps of where this is all going very
0:37:34 quickly.
0:37:40 I’m super optimistic long-term, you know, like, let’s say like 10 years, like super optimistic
0:37:41 about all this.
0:37:45 And I’m also, yeah, I’m also concerned, like the next five years, I think there’ll be a
0:37:46 huge transition.
0:37:49 And most people are not really thinking that through as of right now.
0:37:50 100%.
0:37:51 Yeah.
0:37:52 Selfish question before we go off here.
0:37:57 So my son’s 11 and I always ask people, what should I be teaching him to like, make sure
0:37:58 he can like be successful in AJ?
0:38:00 I mean, you’re going to say it doesn’t even matter because he’s not gonna have a job.
0:38:01 Yeah.
0:38:03 What would you be teaching your son or your child?
0:38:05 Honestly, sales.
0:38:09 I think sales is the one job that’s going to remain forever because I think sales is about
0:38:09 relationships.
0:38:13 And I think people don’t build relationships with AI agents.
0:38:15 They don’t want to be sold to by an AI agent.
0:38:19 So like for that reason alone, I think humans are going to remain in the loop for a very long
0:38:19 time.
0:38:24 I think being a good salesperson is a combination of really solid human skills and like business
0:38:26 skills, which I think is just a powerful combination, period.
0:38:28 So I’m bullish on sales.
0:38:29 Bullish on sales.
0:38:29 Okay.
0:38:29 Interesting.
0:38:32 Yeah, I guess I prepared my son for sales.
0:38:36 You know, he used to be around like parties in San Francisco when he was a little kid and
0:38:38 he got to see like how people would talk about business and stuff.
0:38:41 And I always wondered if that would have some impact on him.
0:38:44 And yesterday, he’s 11 and some of the stuff he’s talked to me about with business is just
0:38:48 mind blowing that he’s already thinking about, you know, the different intricacies of how
0:38:49 to do business.
0:38:50 Last question.
0:38:53 So imagine you have a time machine flow, okay?
0:38:58 And you go to 2050, you step out in San Francisco, what’s different?
0:39:00 Well, assuming we all survive.
0:39:00 Whoa.
0:39:01 Okay.
0:39:03 I mean, look, you know, I mean, that’s what I mean.
0:39:05 I guess I mean, like people should forget.
0:39:06 Assuming we all survive.
0:39:12 Like it’s really hard, almost definitionally, to forecast what happens after the singularity.
0:39:17 Like what’s the name of this sci-fi author who wrote A Fire Open is the Deep, like Werner
0:39:18 Hinge or Hinge or something.
0:39:24 He spent his career writing about a post-AGI, post-singularity world.
0:39:28 And he ended up his career frustrated because he was like, every time I hit a wall, after
0:39:34 20 or 30 years thinking about nothing but this, there’s a thick wall that you cannot go over.
0:39:36 You can’t forecast what happens next, you know?
0:39:37 Right.
0:39:39 And so like, look, you know, you can paint multiple pictures.
0:39:45 I think scenario number one is like post-work utopia, we’re just like all of us hanging out.
0:39:50 We’re like so young and handsome and healthy and rich and like there’s no problem in the
0:39:51 world and all of that stuff.
0:39:52 That’s like scenario number one.
0:39:56 Scenario number two is like, well, all of humanity is dead regardless.
0:40:01 And the world is covered with solar panels and GPUs and data centers.
0:40:06 And I think there’s a scenario to be where it’s like humanity is not totally dead while
0:40:07 like in a reservation somewhere.
0:40:09 By the way, I’m laughing because I’m nervous about it.
0:40:11 Yeah.
0:40:12 No, I mean, 100%.
0:40:12 Yeah.
0:40:13 Yeah.
0:40:15 Those are the sort of scenarios I see on the table.
0:40:15 Yeah.
0:40:16 Okay.
0:40:17 Flow has been awesome.
0:40:19 And like, where should people check you out online?
0:40:19 Yeah.
0:40:20 Lindy.ai.
0:40:22 You know, my email is flow at Lindy.ai.
0:40:28 Just hit me up and I am on Twitter or X as Altimo, A-L-T-I-M-O-O.
0:40:29 Awesome.
0:40:30 This has been great.
0:40:31 We’ll have to have you back on sometime.
0:40:31 Yeah.
0:40:32 Thank you so much, Jason.
0:40:33 Yeah.
0:40:33 Thank you.

0:00:03 Welcome to the Next Wave Podcast.
0:00:09 I’m Matt Wolfe, and today’s guest is building what he calls the last piece of software the
0:00:11 world will ever need.
0:00:17 Anton OCK is the CEO of Lovable, a platform that lets anyone build fully working software
0:00:19 just by describing what they want.
0:00:23 No code, no dev team, just your ideas and AI.
0:00:29 In this episode, we talk about what that means for the future of software development, entrepreneurship,
0:00:30 and even AGI.
0:00:37 Anton shows off a live demo of Lovable, shares how kids and solo founders are launching real
0:00:42 businesses in just hours, and breaks down how this tech could level the playing field for
0:00:43 creators everywhere.
0:00:49 If you’re building anything, tools, products, businesses, this conversation might just change
0:00:51 how you think about the whole process.
0:00:54 So let’s dive in with Anton from Lovable.
0:01:02 This episode is brought to you by HubSpot Inbound 2025, a three-day experience at the heart
0:01:07 of San Francisco’s AI and startup scene, happening September 3rd through the 5th.
0:01:13 With speakers like Amy Poehler, Marquise Brownlee, and Dario Amadei, you’ll get tactical breakout
0:01:18 sessions, product reveals, and networking with the people shaping the future of business.
0:01:19 Don’t miss out.
0:01:23 Visit inbound.com forward slash register to get your ticket today.
0:01:27 Hey, Anton.
0:01:28 Thanks for joining us on the show.
0:01:29 How are you doing today?
0:01:30 It’s great to see you, Matt.
0:01:31 I’m good.
0:01:31 Yeah.
0:01:32 Awesome.
0:01:33 Well, I don’t want to waste your time.
0:01:34 Let’s just jump right into it.
0:01:36 And let’s talk about Lovable.
0:01:39 So I’m curious a little bit about your backstory.
0:01:40 What were you doing before Lovable?
0:01:42 How did Lovable come about?
0:01:43 Why did you decide to build it?
0:01:44 Let’s get the background a little bit.
0:01:45 Yeah, sure.
0:01:47 I go back to my childhood.
0:01:52 No, but I was always a kid that picked apart technology and wanted to understand everything.
0:01:57 And then I found this way to create games when I was like 12 and got books in the library
0:01:58 to learn how to code.
0:02:01 I decided to study at university, as everyone did then.
0:02:04 And then I thought, computer science?
0:02:05 No, I’m going to go into physics.
0:02:11 Because that was where all the people who became the most generalist, both in academia and industry,
0:02:13 studied here, where I’m from, in Stockholm, Sweden.
0:02:15 That was amazing.
0:02:19 I took way too many courses in machine learning, computer science, AI, and math.
0:02:24 And then what I realized, I was at this place where they discovered the Higgs boson,
0:02:27 the particle accelerator in CERN, for three months.
0:02:28 I was like, amazing.
0:02:31 There’s 10,000 super smart people here.
0:02:35 But they’re trying to solve a problem that is very hard to solve.
0:02:36 It’s very inelastic.
0:02:38 You don’t have any real-world impact.
0:02:41 And I’m obsessed about impact and making things happen.
0:02:44 Then I understood, I’m not going to be staying on this track in academia.
0:02:48 So I went into building things in the industry.
0:02:51 And for the last 10 years, I’ve been building AI products.
0:02:55 And I’ve been specifically building great teams that build AI products together
0:02:59 at two of the very, very well-known AI startups here from Stockholm.
0:03:03 And then a bit more than one and a half years ago, I decided to start a new company,
0:03:04 which is what I’m building now.
0:03:08 So with Lovable, can you sort of give the elevator pitch?
0:03:10 Like when somebody asks you, what is Lovable?
0:03:11 How do you describe it to them?
0:03:15 It’s a way to take as the prompt an explanation of an application.
0:03:20 And then AI will build that application for you like it was a software engineer and deploy it.
0:03:21 I think it was you.
0:03:25 I think you said at one point that you’re trying to build the last piece of software.
0:03:27 What does that mean?
0:03:28 Yeah, sure.
0:03:32 I can thank you to what spurred us starting the company, Lovable.
0:03:39 Look, this was early 2023 that it was clear to me like this next generation of AI can actually start reason.
0:03:41 And it’s specifically good at writing code.
0:03:48 If you put it into an advanced system where the reasoning engine is used to take the decision on behalf of a human,
0:03:55 then you’re going to be able to build a completely new type of interface to build software products.
0:03:58 And this AI is going to help developers become more productive.
0:04:05 But the more interesting unlock here is for the 99% who never learned how to code.
0:04:09 I’m not sure about you, Matt, but you’ve probably been frustrated by the difficulty finding great software engineers.
0:04:13 And so I was like my mom and everyone just asked me, how do I find a great software engineer?
0:04:18 So this new interface, you talk to an AI and it builds your product.
0:04:20 You build it together with the AI.
0:04:25 It’s going to let the 99% go from zero to one and enable anyone to,
0:04:31 unlock the creativity to build great companies, to just create things, create great software,
0:04:34 and to build businesses on top of it.
0:04:36 So that’s what we’re set out to do.
0:04:37 And the last piece of software is this.
0:04:40 It’s a platform to build software products.
0:04:44 And it’s going to make sure that humans don’t need the right code anymore if they don’t want to.
0:04:52 So with that in mind, what do you think the role of a software developer or software engineer, what does that look like in the future?
0:04:55 I talk to people who ask me this, like, Anton, what should I do?
0:04:57 I’m an engineer.
0:05:06 And I think engineers should just always put on the hat of and seeing themselves as the person who translates a real-world problem into a technical solution.
0:05:11 And that means different things depending on what type of engineer you are, and it changes over time what you do.
0:05:16 And using AI is, of course, going to be a larger, larger part of that translation.
0:05:25 And now I think what happens when you have AI that makes it faster to create software is that there’s going to be just much more software,
0:05:30 and there’s going to be more iteration cycles to make each piece of software very, very good.
0:05:34 And the jargon for that in some tech companies is to make them lovable.
0:05:40 So I think that’s actually the end outcome of lowering the barriers through AI and new platforms like ours,
0:05:45 making it very, very easy to take an idea, write it in, and you get a full working product.
0:05:49 Right, right. I almost see it as like, you know, like with a symphony, right?
0:05:54 It’s almost like taking the people that are playing the instruments and moving them to the conductor, right?
0:05:59 Now everybody can be the conductor, and you’re telling the various instruments what to do now.
0:06:03 Instead of actually being the player of the instrument, you get to become the conductor.
0:06:05 Yeah, it hooks people.
0:06:14 Yeah, and one thing I’ve loved about this sort of new era of like AI software development is I don’t need to necessarily
0:06:22 build a SaaS product that I’m going out and trying to like raise capital on or, you know, sell it on a monthly fee or anything like that.
0:06:28 I can find like little tiny bottlenecks in my business, little like holes of things that I’m like,
0:06:30 all right, that is kind of a pain in the butt.
0:06:31 I don’t like doing that every day.
0:06:35 Let me build a little software for myself that just fixes that for me.
0:06:40 I can just build it for myself and not have to worry about like trying to build a business around it.
0:06:44 And I think that is one of the coolest things that software like this enables, in my opinion.
0:06:47 Yeah, the personal software trend is also very big.
0:06:48 It’s getting larger.
0:06:49 Yeah, yeah.
0:06:56 So let’s go ahead and maybe jump into Lovable and give a quick demo, show people what it’s sort of capable of.
0:07:01 I know you have a project that you kind of already started working on that we can jump in and tweak with.
0:07:01 Sure.
0:07:05 I prepared a project right ahead of a call previously.
0:07:08 And what I wanted to have in that call, I was going to be asking questions.
0:07:10 And you can use it to ask me questions.
0:07:12 It’s like a webinar Q&A app.
0:07:14 So anyone can answer and input the question.
0:07:18 And I’ll just focus first on what happened as I built this out.
0:07:21 So I basically went to Lovable, which looks like this.
0:07:25 And I put in the first prompt, which was mock-up, the webinar question app.
0:07:33 So then I didn’t want it to be a fully working product where it works across devices, just a mock-up.
0:07:36 And then Lovable went ahead and said like, okay, hey, I’m going to build this for you.
0:07:37 I’m going to choose simple design.
0:07:43 And then it tells me like, if I want to get back in functionality, you can use the Superbase to connect.
0:07:45 And it lets me understand that better.
0:07:51 So then what happened was I got this mock-up UI you can see here, but it doesn’t sync across devices.
0:07:55 So if someone opens this website in one place and answers a question, I don’t see it.
0:07:57 So I just asked AI, how do I do that?
0:07:59 And that’s a big part of like, we’re working with AI.
0:08:01 If you don’t understand, you can just ask.
0:08:01 Yeah.
0:08:07 One thing that I like about what you’re showing here too, is that it actually recommended Superbase, right?
0:08:12 Like if you’re trying to develop a software and you don’t need necessarily know much about software,
0:08:15 you might not know that you need this to be connected to a backend database.
0:08:20 So it’s cool to me that it’s going, hey, we could build this for you, but you also need a database.
0:08:21 Here’s what we recommend.
0:08:22 Yeah.
0:08:28 This is a native integration at this point, because most startups and simple projects that are successful,
0:08:30 they start on Superbase.
0:08:31 So it’s a very popular choice.
0:08:36 And what it does is it told me like, okay, yeah, you need to just connect Superbase.
0:08:39 And then what I did, I went up here and said, yeah, connect to a new project.
0:08:40 Now it’s connected.
0:08:48 And then it tells me, okay, now you can go ahead and add AI functionality, login, or just store data.
0:08:50 So I said, add real-time sync as it explained.
0:08:55 And then it says, I’m going to create the data table, like the place to store the questions.
0:08:59 And I’ll change some in the UI to handle it, to connect to the database.
0:09:02 And then I had to approve, like, okay, run this code.
0:09:03 I got an error.
0:09:04 And then I said, okay, I fixed error.
0:09:06 And then it worked.
0:09:08 And so that’s what happened when I built this.
0:09:12 And if I just open this application, I can send it to you as well.
0:09:15 It’s Q&A dot lovable dot app.
0:09:19 If you go in, you ask me a question, you can try to do it live.
0:09:24 Then it should synchronize real-time across any number of devices, which is a really useful,
0:09:25 simple tool.
0:09:26 And no one has to log in.
0:09:28 It’s just like, you go to the, you scan the QR code.
0:09:29 I ask for a QR code afterwards.
0:09:35 And then I can always pick up this app if I want to have like a Q&A session with a company
0:09:36 or with someone I’m presenting to.
0:09:37 So let me see.
0:09:40 I’ve actually got it open on my screen right now.
0:09:40 Let me see.
0:09:42 If I, let’s see, what should I ask?
0:09:44 Let’s see.
0:09:45 Nothing too personal.
0:09:46 Okay.
0:09:47 Let’s see.
0:09:51 What is the coolest app you’ve seen built with lovable?
0:09:52 Let’s see.
0:09:53 Submit question.
0:09:54 Okay.
0:09:54 That worked.
0:09:55 There it is.
0:09:57 So I actually typed that on my screen,
0:10:00 but you’re seeing it on Anton’s screen if you’re watching the video.
0:10:00 Yeah.
0:10:01 Awesome.
0:10:03 So that’s a good question.
0:10:05 What’s the coolest app I’ve seen built?
0:10:08 I think I saw a better version of ChatGPT,
0:10:09 which I liked.
0:10:13 Like it had more keyboard shortcuts and way to like customize each thread.
0:10:19 I really like that because a lot of the innovation right now actually happens on the AI intersection,
0:10:22 the user interface, like the user experience side.
0:10:25 And that was one of the really cool apps I saw.
0:10:27 It’s also fun to see that people are launching,
0:10:30 we built a lovable app for people to launch the things they built.
0:10:31 Oh, cool.
0:10:33 Almost got like a Reddit style upvote.
0:10:35 And then you get like for projects that I’ve been around,
0:10:38 there’s lots of people getting users through this.
0:10:39 And I haven’t seen all of them,
0:10:41 but there’s so many cool things that people build.
0:10:44 That was just like me demoing how lovable works.
0:10:47 What I could do next is to just show you like how the AI handles a change.
0:10:50 But if I want something to happen instantly,
0:10:51 because the AI is not instant,
0:10:56 I could show you that something you can do here is that you can edit text and style
0:10:58 by just selecting it similar to a website editor.
0:11:01 So you don’t need to wait for the AI to make little sort of changes.
0:11:02 Yeah, for small changes, exactly.
0:11:04 Exactly, yeah.
0:11:05 But let me first ask you something.
0:11:08 Do you have any style you really like to see this in?
0:11:13 I mean, I typically like my websites in dark mode,
0:11:16 and I always like to use like blues and purples,
0:11:17 like my background.
0:11:23 So we will say something about like,
0:11:29 and a cool hacker font and just make it look better.
0:11:33 And what I sometimes do is I just attach a screenshot to the AI.
0:11:35 I pasted it in here for it to see like,
0:11:36 how does it look now?
0:11:38 Let’s make it look more beautiful.
0:11:40 But this is just how lovable works.
0:11:42 You can do some more things,
0:11:44 like you can customize knowledge.
0:11:45 if you wanted to remember,
0:11:49 like always use this way of connecting to an API
0:11:51 that you wanted to use for some type of integration.
0:11:54 If you want your engineer colleague to edit it,
0:11:57 it’s all kept two-way synchronized for them to go to GitHub,
0:12:00 which is like how engineers build software to date.
0:12:00 Right.
0:12:02 And you can invite collaborators,
0:12:04 so I could send you this link,
0:12:08 and then you’ll be able to edit this project if you want to.
0:12:09 That’s most of it.
0:12:11 We’re waiting for the AI to run the change here.
0:12:14 And the last part that a lot of people are missing do here,
0:12:15 now is to add the custom domain,
0:12:16 which you might want to host.
0:12:20 You might want to buy a website domain for your project.
0:12:23 And I think you can even do that inside of Lovable now.
0:12:25 So does Lovable host the whole thing,
0:12:27 and then you’re just sort of pointing your domain name to Lovable?
0:12:28 Yeah.
0:12:29 I can just select the domain here.
0:12:31 I don’t think this domain is free.
0:12:33 I’ll pay inside of this flow.
0:12:34 And then it’s all hosted.
0:12:37 Like we’re using state-of-the-art hosting infrastructure.
0:12:40 Yeah, but I imagine if somebody did want to export all the code
0:12:42 and bring it over to their own host or whatever,
0:12:43 they could do that as well.
0:12:44 Yeah.
0:12:45 So it’s all super flexible.
0:12:49 You can do anything that a human engineer would like to do with this setup.
0:12:49 Oh, nice.
0:12:52 Here’s our new style.
0:12:54 I’m not sure I love it, but…
0:12:55 Hey, put it in what you asked.
0:12:57 Yes, that’s true.
0:12:58 That’s super cool.
0:12:59 Yeah, so that’s it.
0:13:06 The Hustle Daily Show, hosted by John Wigel, Juliet Bennett, Ryla, and Mark Dent,
0:13:09 is brought to you by the HubSpot Podcast Network,
0:13:11 the audio destination for business professionals.
0:13:15 The Hustle Daily Show brings you a healthy dose of irreverent,
0:13:18 offbeat, and informative takes on business and tech news.
0:13:23 They recently had an episode about advertisers wanting billboards in space.
0:13:26 It was a really fun and informative episode.
0:13:27 I suggest you check it out.
0:13:30 Listen to The Hustle Daily Show wherever you get your podcasts.
0:13:37 So one of the things, like, when it comes to, you know,
0:13:40 using AI for code that I’ve ran into a few times,
0:13:44 is like, I’ll have it build something, and then there’ll be a bug,
0:13:46 and then I’ll say, hey, this bug is popping up.
0:13:47 Can you fix it?
0:13:50 It’ll fix that bug, and then maybe introduce a new bug.
0:13:55 Or it’ll keep on, like, having that same bug over and over and over again.
0:13:59 I know a lot of the LLMs have gotten better and better and better over time.
0:14:00 We’ve now got Cloud 4.
0:14:02 We’ve got Gemini 2.5 Pro.
0:14:04 A lot of these LLMs have gotten a lot better at coding.
0:14:07 But I’m curious, like, how does Lovable specifically
0:14:10 maybe help overcome some of that kind of frustration
0:14:12 that, like, the Vibe coders might have?
0:14:13 Yeah.
0:14:17 So Lovable is not just a call to Cloud, like the new Cloud model.
0:14:22 It does a few agentic chain, which is that it tries to understand,
0:14:23 okay, what’s the context here?
0:14:27 Like, exactly what information is most relevant
0:14:30 to solve this specific problem that you’re having.
0:14:32 If you’re seeing a repeated bug, like, that’s one type of situation.
0:14:38 And then we’re applying best practices that we’ve been iterated ourselves towards
0:14:41 to solve that specific type of context that you’re in.
0:14:43 Okay, you’re stuck with the same type of bug.
0:14:45 And we feed in some of those best practices
0:14:48 that are, like, adopted to work
0:14:53 for the specific technology stack that Lovable applications are built on.
0:14:54 So that’s what we do to date.
0:14:56 And another important thing is that we have to give access
0:15:00 to, like, what human engineers use to debug,
0:15:05 which is the AI is able to read all the error messages
0:15:08 and, like, all the logs that are created
0:15:10 as the user interacts with the website.
0:15:12 So that’s fed into the AI system.
0:15:14 So then you can see that if there’s a bug,
0:15:16 it can really get much more of a picture of, like,
0:15:17 what actually happened here,
0:15:21 and then use that in terms of figuring out the error.
0:15:23 That’s what takes most time for most software engineers
0:15:25 to understand what is it exactly that goes wrong.
0:15:26 It doesn’t work.
0:15:30 It’s not sufficient information to fix it.
0:15:31 So those are some of the things we’re doing,
0:15:33 and we’re working on a lot more.
0:15:33 Very cool.
0:15:35 This is sort of getting more into the, like,
0:15:37 theoretical, philosophical kind of area.
0:15:40 And I’m not sure if this is something you’ve thought about or not.
0:15:43 So if you haven’t, we can just always skip over it.
0:15:47 But I’m curious, like, when everybody has access
0:15:50 to be able to develop any software,
0:15:54 how do companies, like, actually build a moat?
0:15:56 Like, have you thought about that at all?
0:15:59 Like, how would a software company actually build something
0:16:01 if, like, anybody else can just go and use a tool
0:16:02 to build the same thing?
0:16:04 Like, how do businesses get formed around this kind of thing?
0:16:08 I mean, you don’t need a moat to build a great business.
0:16:08 Right.
0:16:09 I don’t think so.
0:16:13 And, I mean, most of the moats are the same.
0:16:18 I think there is a bit of a moat in terms of just trust
0:16:19 and knowing who’s behind this,
0:16:23 that these people who built this have my best intentions at heart.
0:16:23 Right.
0:16:25 That will always remain.
0:16:28 Then there are, like, network effects moats.
0:16:31 So I guess, like, one on that all your friends are using it,
0:16:32 and then it becomes more productive to use this tool.
0:16:36 And one on that this tool is connected to everything else out there,
0:16:37 and that’s, like, a network platform effect.
0:16:38 Right.
0:16:42 And I think, like, maybe at some point there’s an economy of scale as well,
0:16:45 like, that you can make it give a better value proposition
0:16:45 because you’re larger.
0:16:49 That hasn’t shown to be so useful in software businesses,
0:16:51 but maybe that’s going to be the case in the future as well.
0:16:52 Yeah, yeah, yeah.
0:16:55 So I want to shift the conversation quickly to the topic of AGI,
0:16:57 because I know in the past you’ve also mentioned
0:17:00 that you want to help contribute to getting to AGI.
0:17:03 So a couple questions there.
0:17:04 A, how would you define AGI?
0:17:08 Because I feel like, obviously, everybody kind of has a slightly different definition.
0:17:11 And then the follow-up to that is, like, how does Lovable,
0:17:14 or what you’re trying to build, sort of get us closer to that?
0:17:15 Yeah.
0:17:23 My favorite definition is that at the point when anything you could hire a human remote
0:17:27 worker for can be done with AI, that’s when you have AGI.
0:17:32 It hedges you against that humans can do some things that only humans can,
0:17:34 because we humans don’t want to talk to a machine.
0:17:35 We want to talk to a human.
0:17:39 But if it’s a remote, like a remote and over Slack, I think,
0:17:41 then it’s only cognitive labor.
0:17:42 It’s a pure cognitive labor.
0:17:45 So I think that’s a pretty clear way to define,
0:17:50 is it intelligence and not just a human that has networks and connections with other humans?
0:17:51 Right.
0:17:53 But how do we get there?
0:18:00 I think building systems that execute, write and execute code is a huge part of this.
0:18:08 And increasingly, what I’m thinking is that we are not going to work on the foundation model layer.
0:18:14 We’re creating the most delightful and intuitive interface to interface with this type of technology.
0:18:21 And now it’s for spinning up software, and that’s hosted, available for anyone.
0:18:26 In the future, we’re naturally adding a lot more types of interactions into our interface,
0:18:29 like, oh, browse the web for me and find all these interfaces, these things,
0:18:34 and then put them into a website, or even like, okay, can you check for all the feature requests
0:18:36 from my email and then implement some of them?
0:18:41 Like, that’s the type of direction that we would go for on the very long term.
0:18:47 And that type of interface is going to be one of the most important things in how humans perceive
0:18:49 the level that AI has reached, right?
0:18:53 Like, it’s a good friend of mine told me way, way back, 10 years ago,
0:18:55 like, the AI is never better than the UI.
0:18:59 So if you can’t, as a human, get value from it, then it’s worthless, right?
0:18:59 Yeah.
0:19:04 Do you think that, like, the future of UI, the future of user interface is going to be
0:19:10 what it is now, where people are sort of typing prompts, and we have these sort of visual user
0:19:13 interfaces, or do you think it’s going to switch to some alternate modality?
0:19:17 No, I think it’s going to be pure mind reading in the future, right?
0:19:19 We don’t know what it’s going to be yet.
0:19:25 It’s going to be a combination of things, like, us humans are really good at getting a lot
0:19:31 of information visually, so that’s going to continue to be a big part, and like, we’re
0:19:35 not as good at getting a lot of information by reading text, I think, as like, just looking
0:19:37 at a picture and boom, you get a lot of information really fast.
0:19:38 Right.
0:19:42 That’s going to be a part of it, and then, like, how you as a human communicate as much
0:19:46 information as possible to an AI, which is going to be important as well.
0:19:51 I mean, at some point, I do think we’re going to see more and more adoption of, like, brain
0:19:57 computer interfaces, but just speaking or lip reading might be like an emerging pattern
0:19:58 UX-wise with AI.
0:19:59 Yeah, yeah.
0:20:03 You know, I had some chats with people over at, like, Microsoft and Google and that sort
0:20:10 of thing, and sort of their position is that they want AI to be much more predictive, right?
0:20:14 Like, it knows what you want to do before you ask it to do the thing.
0:20:18 It’s going to get to this point where it starts to understand you, it understands your patterns,
0:20:23 it understands what you do on a daily basis, anticipates it, and then just gets ahead of
0:20:23 you on it.
0:20:28 So, I don’t know, to me, that’s like a real sort of fascinating future that we’re heading
0:20:29 into with AI.
0:20:30 Yeah, that’s huge.
0:20:35 Is there anything else that we didn’t cover about, you know, Lovable and what you’ve been
0:20:37 building that you think we should be covering?
0:20:42 I could talk a bit about, like, where this technology is giving the most value today.
0:20:48 We spoke to one of our users recently, Felipe, I think it was a very fun story where he had
0:20:50 built large companies before.
0:20:55 He raised $50 million, hired 130 engineers, and now he’s past that.
0:20:59 And he’s just building a business himself using Lovable.
0:21:04 And then he can take all the ideas instantly and it moves so much faster, which is a bit
0:21:08 paradoxical, to having this large organization where there are many chains of communication.
0:21:12 And it sounds very productive to have 130 engineers, right?
0:21:17 But he’s making tens of thousands of dollars on this, like, small new business that he’s
0:21:18 growing organically.
0:21:23 This is like the stereotypical AI native founder that I think we’re going to hear a lot more
0:21:27 from in how one person can build much faster than larger companies.
0:21:33 And we’re doing where they let AI do more and more, you know, the building part, but also
0:21:34 the marketing side.
0:21:37 And all of that is going to be like one human and a lot of AI systems.
0:21:39 So that’s what we’re seeing.
0:21:43 What also inspires me a lot is that kids love to use Lovable because they are super creative,
0:21:43 right?
0:21:45 They love to create things.
0:21:49 And I’ve seen many, like, 14-year-olds and even younger who post, like, they’re selling
0:21:54 something online or they’re, like, services to walk the dogs with a website built on Lovable.
0:22:01 And increasingly now, since we launched a Teams plan recently, Lovable is getting huge in
0:22:05 larger companies, like, Fortune 500 companies that are individuals in Teams.
0:22:11 They accelerate how the team and the entire company takes decisions, both in the terms of,
0:22:12 like, okay, we should really build this thing.
0:22:13 Look, I’ve already built it.
0:22:14 It’s working.
0:22:20 And then they look in engineering and for building tools that just accelerates, like, finance,
0:22:23 the marketing, building landing pages and all of that.
0:22:26 So it’s fun that it’s a tool that’s being used for so many different things.
0:22:28 And we’re just keeping up on our side.
0:22:33 Yeah, I think last year at some point, Sam Altman from OpenAI mentioned that he thinks
0:22:37 within the next couple of years, we’re going to see the first billion-dollar company built
0:22:38 by one person, right?
0:22:43 And then Daria from Anthropic just said it again, like, two weeks ago that he thinks within
0:22:50 2026, we’ll probably see the first one-person billion-dollar company, which is absolutely wild.
0:22:52 Also, you mentioned kids are building apps.
0:22:57 I actually had a conversation with Kevin Scott, the CTO of Microsoft, and he told this whole
0:23:01 story about how his daughter built an entire app for her school.
0:23:05 And he was frustrated because she didn’t even consult with him.
0:23:06 And he’s a software developer.
0:23:08 She just went and built it herself.
0:23:12 I mean, what you’re saying, we’re definitely seeing more and more of.
0:23:18 It’s fun that kids are going to be, of course, better than older people generally to use AI.
0:23:22 And it’s going to be such a difference in how productive you are if you’re good at using
0:23:22 AI.
0:23:23 Yeah, yeah.
0:23:26 I have one small rabbit hole I want to go down with you really, really quickly.
0:23:32 I’m curious how the developer community as a whole has received something like Lovable.
0:23:36 Because I know some developers probably absolutely love it.
0:23:37 It speeds up their time.
0:23:43 But then there’s also that sort of existential fear that their skill that they’ve been building
0:23:43 is no longer needed.
0:23:45 How has the reception been so far?
0:23:50 If you just look at the product, the features, a lot of developers love that you just create
0:23:51 a fully working application.
0:23:57 And then if you want to go in and customize, use your normal ID, you just sync it with the
0:23:58 like secure code base.
0:24:04 And then you’re getting more done and you’re shipping more value to your customer, your employer.
0:24:07 And that’s like a very positive reception generally.
0:24:10 If you do zoom out and you’re like, oh, but wait, where is this actually headed?
0:24:11 Yeah.
0:24:18 Then it is the case that people are like, wait, what’s my role in all of this areas?
0:24:24 But I think it’s not so different from anyone working in white collar jobs that everything
0:24:26 is going to be easier and easier to automate.
0:24:31 If you’re on top and master these tools, you’re going to be much, much more valuable in the
0:24:32 workplace.
0:24:37 But otherwise, you’re not going to maybe have an as cushy job as a software engineer
0:24:37 has been.
0:24:41 But you’re going to have to combine that with like doing sales or doing something more manual
0:24:42 as well.
0:24:47 There’s going to be a potentially the long term reduction in how many people sit and build
0:24:47 software.
0:24:48 Yeah.
0:24:49 No, I couldn’t agree more.
0:24:53 And like you mentioned, a lot of the white collar work is in that same boat, right?
0:24:57 Like if your job is sitting around looking at Excel spreadsheets all day or bookkeeping
0:25:04 or doing research for a law firm, a lot of that work is also going to probably get automated
0:25:07 away through AI fairly quickly too.
0:25:12 And just one thing on the software, the demand for software doesn’t end.
0:25:17 Like there’s seems to be a lot of things that can improve, be improved with software.
0:25:22 And hence, there’s going to be more people building software, maybe fewer people writing code.
0:25:23 That’s how I see it.
0:25:24 Oh yeah.
0:25:25 That makes a lot of sense to me.
0:25:28 Well, Anton, this has been absolutely amazing.
0:25:32 I know you have to get off to another meeting, so I don’t want to waste any more of your time.
0:25:34 So the app is over at lovable.app.
0:25:36 That’s the best place to go to get it?
0:25:36 Yeah.
0:25:38 Or lovable.dev is the normal one.
0:25:39 Oh, lovable.dev.
0:25:39 Okay.
0:25:41 So head over to lovable.dev.
0:25:45 Is there any place that you maybe want people to follow you on social media or anything like
0:25:46 that after listening to this interview?
0:25:53 I share fun takes on building from Europe and on the AI space and what’s happening with
0:25:55 our advancements at my Twitter.
0:25:57 That’s my first and last name combined.
0:25:58 Awesome.
0:26:01 Well, thank you so much for hanging out with me and having this conversation.
0:26:06 It’s been really fun and, you know, really excited to see how lovable evolves over time.
0:26:07 So really appreciate it.
0:26:08 Thank you.
0:26:08 Likewise.
0:26:09 It was a pleasure.
0:26:11 I’m looking forward to another chat in the future.
0:26:12 Absolutely.
0:26:12 Absolutely.
0:26:12 Absolutely.
0:26:12 Absolutely.
0:26:27 We’ve got a major announcement.
0:26:32 HubSpot is the first CRM to launch a deep research connector with ChatGPT.
0:26:37 Customers can now bring their customer context into the HubSpot deep research connector and
0:26:39 take action on those insights.
0:26:43 Now you can do truly remarkable things for your business.
0:26:48 Customer success teams can quickly surface inactive companies, identify expansion opportunities
0:26:51 and receive targeted plays to re-engage pipelines.
0:26:56 Then take those actions in the customer success workspace in HubSpot to drive retention.
0:27:02 Support teams can analyze seasonal patterns and ticket volume by category to forecast staffing
0:27:08 needs for the upcoming quarter and activate Breeze customer agents to handle spikes in support
0:27:08 tickets.
0:27:11 This truly is a game changer for the first time ever.
0:27:17 Get the power of ChatGPT fueled by your CRM data with no complex setup.
0:27:23 The HubSpot deep research connector will automatically be available to all HubSpot accounts across
0:27:27 all tiers that have a ChatGPT team, enterprise, or Edu subscription.
0:27:33 Turn on the HubSpot deep research connector in ChatGPT to get powerful PhD level insights from
0:27:34 your customer data.
0:27:36 Now let’s get back to the show.
0:27:39 Thank you.

0:00:07 What if there was an AI doctor that knew everything about you, your sleep, your history, your
0:00:11 habits, and could give you better advice than any human doctor ever could?
0:00:17 Today, I’m talking with Max Marchionne, founder of Superpower, a company using AI to help
0:00:20 people live longer, healthier lives.
0:00:22 It was a fascinating conversation.
0:00:26 We talked about how AI is changing medicine, how to actually live to 120 or, I don’t know,
0:00:32 maybe 200, and why Max thinks things like supplements are overrated, but he still takes 30 a day.
0:00:38 So if you care about living longer and staying healthy in the age of AI, this is the episode
0:00:39 for you.
0:00:47 Cutting your cell cycle in half sounds pretty impossible, but that’s exactly what Sandler
0:00:48 Training did with HubSpot.
0:00:53 They used Breeze, HubSpot’s AI tools, to tailor every customer interaction without losing their
0:00:54 personal touch.
0:00:56 And the results were pretty incredible.
0:00:59 Click-through rates jumped 25%.
0:01:03 And get this, qualified leads quadrupled.
0:01:04 Who doesn’t want that?
0:01:07 People spent three times longer on their landing pages.
0:01:08 It’s incredible.
0:01:13 Go to HubSpot.com to see how Breeze can help your business grow.
0:01:18 Max, thanks for coming on the show.
0:01:20 Nathan, thank you for having me here.
0:01:21 Yeah.
0:01:24 So this is kind of a, you know, unconventional episode for us.
0:01:28 Part of my motivation for starting this podcast in the first place was to help people thrive
0:01:30 in the age of AI and, you know, we’re going through this huge transitionary period.
0:01:35 Normally, we cover that more in like a business context of how you can use AI and help, you
0:01:37 know, help you in your work or how to get ahead in business.
0:01:40 But also, I feel like part of that is you also have to be healthy to actually live to see all
0:01:41 these things that are happening, right?
0:01:46 And I saw you guys post on Twitter when you first announced superpower and thought it was
0:01:50 fascinating that you guys are like actually using AI to help people know how to live a better
0:01:51 life and how to be healthier.
0:01:52 Is that kind of the general gist of it?
0:01:53 Totally.
0:01:58 So it’s my belief that every single person on earth will have an AI doctor and that doctor
0:02:02 will be so much better than any human doctor was ever able to be, right?
0:02:02 Right.
0:02:04 We won’t just replicate the best of medicine.
0:02:10 We’re actually going to enter into the golden age of medicine because now the millions of data
0:02:12 points about you can be processed.
0:02:16 The millions or thousands of papers about medicine can be processed.
0:02:17 We can discover new things.
0:02:20 We can start to detect things years in advance.
0:02:21 We can start to personalize medicine.
0:02:26 So all of these things, which were lofty goals, are now on the foreseeable horizon.
0:02:27 That’s awesome.
0:02:29 I’ve been interested in longevity for a long time.
0:02:32 Like I watched Andrew Huberman before that, Tim Ferriss.
0:02:35 I’ve met Aubrey de Grey in San Francisco before.
0:02:38 But honestly, I’ve just never really done much to track my own health.
0:02:40 Like besides like, oh, I try to eat healthy.
0:02:41 I take vitamin D.
0:02:43 I go for walks and exercise.
0:02:46 But I’ve never done any of the tracking or the sleep tracking stuff or anything.
0:02:50 Like my understanding is you’ve been tracking yourself since you were like 16 or so.
0:02:51 Is that correct?
0:02:52 Yes.
0:02:56 I went through a long period of misdiagnosis and tested and tracked everything.
0:02:59 It would take me three hours to get to sleep every night.
0:03:00 And I had no idea what was wrong with me.
0:03:04 And as you do, you start trying to learn.
0:03:06 So I remember wearing a big Gen 1 aura ring.
0:03:08 This was when it had just come out.
0:03:10 People used to make fun of me for tracking sleep.
0:03:11 You didn’t do it back then.
0:03:16 And I, every single day, wake up and score my sleep qualitatively.
0:03:16 How did I feel?
0:03:17 Yeah.
0:03:20 Then I score my sleep at the end of my day at the end and say, how did I feel?
0:03:24 And then I look at the quantitative data from the ring and from several other wearables to
0:03:26 see what was showing up in the quantitative data.
0:03:28 And then I said, here were all of the different things I did.
0:03:30 Here’s whether I drunk alcohol or not.
0:03:32 Here is the temperature in the room.
0:03:34 Whether I exercise when I last drank caffeine.
0:03:35 How much caffeine.
0:03:38 And I started to learn what works and what doesn’t.
0:03:42 And you start to get a really good sense of what works and what doesn’t.
0:03:43 What have you learned?
0:03:44 This is partially a selfish episode too.
0:03:46 I’m trying to learn as much to live as long as possible.
0:03:48 So all the obvious things we already know.
0:03:50 Exercise helps a cold room.
0:03:52 Helps a dark and quiet room.
0:03:53 Oh, I’m not going to bore you with that.
0:03:54 I’ll say the interesting things.
0:04:01 So half a glass of red wine, for me, three hours before bed is better than none if we’re
0:04:03 just looking at qualitative and quantitative sleep.
0:04:05 Again, that’s my personal experience.
0:04:06 It might not be true for everyone.
0:04:11 Of the different supplements, Pharmagabba is probably the most effective at increasing HRV.
0:04:14 Very few of the others increase HRV.
0:04:20 Doing either heart rate variability training before bed or meditation before bed has an enormous
0:04:27 impact for me, partially in quieting the mind and partially in lowering resting heart rate.
0:04:29 Lowering resting heart rate matters a lot.
0:04:35 If you want to force your heart rate to lower, do 10 seconds in, 20 seconds hold, 20 seconds out.
0:04:41 Like extreme, but that extreme hold, extreme outs will lower your resting heart rate, I find,
0:04:42 faster than other things.
0:04:44 That’s kind of like what Wim Hof does, right?
0:04:46 But in terms of like holding it in and letting it out?
0:04:48 His is a little bit different.
0:04:48 Is it?
0:04:54 Wims is you breathe vigorously in and out 30 times over, and then you exhale, and then
0:04:55 you hold your breath.
0:04:55 Right.
0:04:58 So that will increase resting heart rate rather than decrease it.
0:05:00 Okay, so it’s the opposite of what you want.
0:05:00 Okay.
0:05:00 Yeah.
0:05:01 Cool.
0:05:03 Yeah, I’ve actually tried to meditate in the morning.
0:05:05 I used to do it at night, and I stopped.
0:05:07 I’m not sure why I stopped.
0:05:11 I remember when I would do it, I would have like the best sleep ever, and I’d just instantly
0:05:14 go to sleep, you know, because otherwise I lay in bed and I’m just thinking about a thousand
0:05:15 different things.
0:05:16 It seemed like it really helped for that.
0:05:17 Yep.
0:05:18 A hundred percent.
0:05:18 Cool.
0:05:22 So my understanding is you had like a cool demo to show me of your company’s superpower.
0:05:23 Like, I’d love to see it.
0:05:24 Yeah, let’s do it.
0:05:26 I’ll share how the membership works today.
0:05:31 The idea is we want to create the one place people come to, to take control of their health.
0:05:34 And there’s three parts to our annual membership today.
0:05:38 Part number one is collecting as much data on someone as possible.
0:05:44 So at the start of the year, we’ll send a nurse to someone’s home, collect over a hundred plus
0:05:47 blood biomarkers, and then allow them to get all of these other tests as well.
0:05:50 Gut microbiome, environmental toxins, cancer screening.
0:05:56 We’ll pull all of that in to a data page where you can visualize all of your past results over
0:05:57 time as well.
0:05:59 We have these beautiful graphical ranges.
0:06:04 We’ve defined not just normal and out of range, but also an optimal range, right?
0:06:07 Which we’ve worked with a PhD out of Oxford to define.
0:06:11 And then we’ll pull in all of someone’s past medical records as well, integrating with the
0:06:11 EMRs.
0:06:17 And this is important because having full context on the person is important in AI-led
0:06:18 paradigm of medicine.
0:06:21 And then finally, we’ll integrate someone’s wearables too.
0:06:25 So that’s part number one, collect as much data as someone else as possible, test the whole
0:06:26 body, and people love that experience.
0:06:29 We’ll visualize and we’ll show biological age as well.
0:06:34 Part number two is now that we know so much about people, how do we actually connect the
0:06:35 dots, right?
0:06:38 How do we tell someone, look, here’s the thousands of data points, but here’s what’s really going
0:06:38 on.
0:06:40 Here are the issues you have today.
0:06:41 Here’s what might come up in the future.
0:06:45 So we build some of this, this big protocol, which is a combination of our medical team
0:06:46 and AI.
0:06:51 And people love this because they get to the end of it and they’re like, oh, whoa, now I
0:06:53 really see what’s going on in my body.
0:06:55 I never had any doctor connect the dots like this before.
0:06:58 And we’ll tell someone, here’s exactly what you should do.
0:07:03 And the final part, what I used to find frustrating about medicine is you’d leave the doctor’s office
0:07:05 and they’d say, good luck, you’re left to your own devices.
0:07:09 What we try to do is after we tell you what to do, we try to actually help you do it.
0:07:11 So any follow-up diagnostics are in one place.
0:07:15 At any given point in time, we’ll show you, here’s the exact next step.
0:07:17 You should never have to think about it.
0:07:18 Here’s the exact next step.
0:07:24 If you need access to a supplement or pharmaceutical, we’ve brought only our favorite, highest quality
0:07:28 products into one place, all 20% cheaper than Amazon or elsewhere.
0:07:35 The idea is if you’re a member, it should be cheaper and easier to access the best care within the ecosystem and hunting outside.
0:07:43 And then the final thing is if you have a health question, you can pull out your phone, you can message your health AI or send a text message to your concierge team.
0:07:43 Oh, that’s awesome.
0:07:45 So this is basically like your own AI doctor.
0:07:50 You give it all the data about yourself and then now it knows all it’s about you versus having to go into a doctor.
0:07:51 So this one here is the AI.
0:07:51 Okay.
0:07:52 Yeah.
0:07:53 And then this one here is a medical team.
0:07:57 There’s this three humans on it, assisted with technology and AI.
0:07:57 Right.
0:07:58 So you have both.
0:07:59 So basically that’s what we do today.
0:08:01 Test the whole body twice a year.
0:08:02 We’re a hundred plus labs.
0:08:08 Connect the dots, tell you what to do and make it really, really easy to take action with a lot of things in one place.
0:08:14 And that membership today we’re doing for $42 a month or $499 a year, which I think is like quite affordable.
0:08:18 So we’re growing quite quickly and having a lot of people wanting to sign up.
0:08:19 Yeah, it’s awesome.
0:08:28 I always thought it was so weird that you go into the doctor and you see them for like 15 minutes, 30 minutes, whatever, and they’re on to the next thing and they barely know who you are or anything about you.
0:08:38 They glance at this thing, you know, showing all your data and it seems like this will make it possible where, you know, the AI will actually understand who you are and what’s going on in your own body, like personalized.
0:08:39 How long have you guys been doing this?
0:08:41 Like what was the inspiration for starting it?
0:08:45 I, as I mentioned, went through a challenging personal health experience.
0:08:46 Yeah.
0:08:48 Chronic migraines, chronic sinusitis.
0:08:50 I had surgery, was taught to medicate for life.
0:08:55 Didn’t really know what was going on until I found a genius doctor who connected the dots.
0:09:02 And it made me realize there’s a big gap between the best of healthcare and what most people have access to.
0:09:09 And I’ve believed for the past decade or so that for as long as that gap exists, someone or some company has to come along and close it.
0:09:21 And I didn’t think that was possible until November, 2020, 2022, when ChatGBT was launched and started to show a path towards what was possible with these AI tools.
0:09:30 And we started the company with the belief that we could take the very best of medicine that today costs hundreds of thousands of dollars and make that accessible for a few hundred dollars.
0:09:32 So that was the initial impetus.
0:09:32 That’s awesome.
0:09:37 I think I have how many people that a lot of times with doctors, they’re not like checking for cancer or things like that ahead of time.
0:09:39 They’re just like telling you like, oh, you’ve got cancer.
0:09:41 They’re not doing a lot of preventive stuff.
0:09:43 And yeah, so it’s just like they’re going to be a game changer.
0:09:45 Yep, 100%.
0:09:48 How can people listening to this, how can they actually like live a longer life?
0:09:53 Just use superpower or is there any other like practical advice you can give people that they could just get started with today?
0:09:57 Oh man, if you want to live a longer life, we’re going to need biotech.
0:10:00 And biotech might be expensive.
0:10:03 It’s the unfortunate reality of where the future is going.
0:10:07 But biotech will define the future more than healthy eating.
0:10:10 And access to biotech might be unevenly distributed.
0:10:14 And insofar as AI can make us more productive, that’s interesting.
0:10:21 I think today, these foundational models, if it has full context, they’re very good at connecting the dots.
0:10:23 They’re very good at working out what’s going on.
0:10:26 There are ways to prompt it to make it more effective.
0:10:28 I find the reasoning models like O3 are the best.
0:10:33 I’ll prompt it with something like, you’re an elite integrative medicine doctor.
0:10:34 You’re the best in the world.
0:10:36 You know everything about medicine.
0:10:37 You know everything about me.
0:10:39 You know everything at the frontier.
0:10:42 You also understand traditional Chinese medicine and Ayurvedic medicine.
0:10:44 You’ve seen everything.
0:10:46 You think deeply about the root cause.
0:10:48 And my doctor, here’s what’s going on for me.
0:10:49 Explain everything.
0:10:50 I’ll give it everything.
0:10:53 My upbringing, my life, what I eat, what I do.
0:10:55 And if you give it all of that, it’s really damn good.
0:10:57 So that’s just ChatGPT.
0:11:01 Now, there’s a lot of other things you can do with prompting that make it far better.
0:11:04 We try to do a lot of that with the superpower AI.
0:11:06 But you can do a lot now with ChatGPT.
0:11:07 They’re pretty impressive.
0:11:08 Yeah, I’ve been surprised.
0:11:12 I had a hand injury and it was like lingering for like three or four months.
0:11:15 And I started chatting with ChatGPT about it.
0:11:17 And like, it told me exactly what was going on.
0:11:19 Now, I went to a doctor and they told me basically the same thing.
0:11:22 And so I’m like, oh, in theory, I could have just used ChatGPT.
0:11:24 I didn’t necessarily have to go to a doctor.
0:11:24 Oh, yeah.
0:11:27 These AIs are better than the doctors often now.
0:11:34 There was this paper that came out that showed doctor accuracy, doctor plus AI accuracy, and then just AI accuracy.
0:11:38 The just AI was better than the doctor plus the AI.
0:11:40 It was more accurate.
0:11:47 And that’s the model today with limited context, without being specialized or set up for medicine-specific use cases.
0:11:51 Where we’re going is models that are more specialized with more context.
0:11:51 Yeah.
0:11:56 We’re still holding on to humans as the practitioners of the art of medicine.
0:12:02 And I think anyone who does that is still reasoning from the past and reasoning from the current point in time we’re at.
0:12:06 Whereas I think what we need to consider is the rate of change and the directionality of change.
0:12:11 And if we think models today are impressive, well, in two years’ time, we’re going to be blown away.
0:12:12 Right.
0:12:12 Yeah.
0:12:18 I mean, if you think about, you know, doctors, they get paid so much because they go to school for a really long time and they remember lots of stuff.
0:12:24 And, you know, they read so many different books and they probably retain, like, I don’t know, 5% of the knowledge or something like that that’s actually out there.
0:12:27 With AI, obviously, they can, like, know every single paper that’s out there.
0:12:29 And then, like you said, even other stuff, you know, Chinese medicine, whatever.
0:12:34 You can look at the collective knowledge of humanity, you know, with health and actually apply that.
0:12:35 It’s a fascinating concept.
0:12:36 Yeah.
0:12:40 I think that’s such a good point, which is that a large part of medical school is remembering knowledge.
0:12:41 Right.
0:12:46 And one of the magical things about technology is we don’t have to remember knowledge anymore.
0:12:48 So, yeah, look, I tend to agree.
0:12:50 And we already implicitly know this is true.
0:12:53 There are people who hear this and be like, what do you mean he’s talking?
0:12:55 I’m like, we already implicitly know this is true.
0:12:58 When you have a health question, what do you do?
0:13:00 You go to Google or ChatGPT before the doctor.
0:13:01 Typically what people do.
0:13:04 Then you go to the doctor and they might say some things.
0:13:08 And then you go back to Google and ChatGPT after the doctor.
0:13:09 And you know what?
0:13:12 During the consult, the doctor’s using Google and ChatGPT as well.
0:13:17 We already implicitly know that people go to the algorithm a lot of the time before they go to the doctor.
0:13:18 Right.
0:13:21 It is like people are still going to want to see a doctor in a lot of cases, right?
0:13:24 They want to talk to a person and this will probably change over time.
0:13:27 I think as of right now, people want to talk to a doctor and be reassured, things like this.
0:13:30 And I guess also if you get like a bad diagnosis or something too, right?
0:13:34 Like that’s probably where you would want a human telling you, I guess, versus an AI.
0:13:36 But in the future, that will probably change.
0:13:37 Maybe in the future, it’s entirely AI.
0:13:40 You know, maybe it’s all superpower or other companies like superpower, right?
0:13:42 Today, 100%.
0:13:47 People, trust, humans, the opinion of a doctor validates or invalidates an AI.
0:13:50 It makes you more likely to follow a recommendation.
0:13:52 That’s all going to change.
0:13:53 It’s inevitable.
0:13:55 I’ll give you maybe a hypothetical.
0:13:57 Let’s say you’re boarding an airplane.
0:13:59 You’re going to do a 15-hour trip.
0:14:04 And over the loudspeaker, you hear, the autopilot and the technology is not working this flight.
0:14:06 The pilot’s just going to fly it.
0:14:09 You’re going to be like, get me out of here.
0:14:10 Where’s the parachute?
0:14:15 We’re sorry to tell you, but the AI and the technology is broken.
0:14:16 The pilot is going to be flying the rest.
0:14:19 You’re like, no, no, this is like, are you fearing for your life?
0:14:21 Especially if you’re going to turbulence or something, right?
0:14:22 You’d be like freaking out.
0:14:24 That’s where we’re going to get to, right?
0:14:24 Yeah.
0:14:30 And if you said to someone in 1940 that the AI is going to fly the plane, they would have been like,
0:14:31 no way, that’s so unsafe.
0:14:32 I need the human there.
0:14:34 Well, now it’s the opposite.
0:14:34 Yeah.
0:14:37 I think we’re going to see that now we need the AI there.
0:14:38 We’re going to see the same in healthcare.
0:14:40 Again, we’ll like deny it and then it will happen.
0:14:41 And yeah.
0:14:42 Yeah.
0:14:42 Yeah.
0:14:45 It feels like younger people embrace it, you know, faster.
0:14:49 I talked to people in my family recently who didn’t even know that AI helps fly the planes.
0:14:53 I think there’s actually a lot of people who don’t or aren’t even aware of that, actually.
0:14:54 Totally.
0:14:55 There’s a reason why there’s still a pilot in the cockpit.
0:14:56 Right.
0:14:58 The pilot maybe does a couple of things.
0:15:01 But if you chat with a commercial airline pilot, ask them about how much they actually do.
0:15:02 Right.
0:15:02 It’s not much.
0:15:03 Yeah.
0:15:04 I live here in Japan.
0:15:04 It’s even with the trains.
0:15:05 You notice that too.
0:15:06 I’m pretty sure most of it’s automated.
0:15:09 And you see the guy up there and he’s like, you’re just really casual.
0:15:12 He’s like sipping a coffee, not really paying much attention to anything.
0:15:15 He kind of makes sure there’s nothing on the track or like, you know, make sure people are
0:15:16 okay and stuff.
0:15:19 But it seems like it’s all just kind of ran by computer systems.
0:15:25 Hey, we’ll be right back to the show.
0:15:28 But first, I’m going to tell you about another podcast I know you’re going to love.
0:15:30 It’s called Marketing Against the Grain.
0:15:33 It’s hosted by Kip Bodner and Kieran Flanagan.
0:15:38 And it’s brought to you by the HubSpot Podcast Network, the audio destination for business
0:15:38 professionals.
0:15:43 If you want to know what’s happening now in marketing, especially how to use AI marketing,
0:15:44 this is the podcast for you.
0:15:50 Kip and Kieran share their marketing expertise, unfiltered in the details, the truth, and like
0:15:51 nobody else will tell it to you.
0:15:58 They recently had a great episode called Using ChatTBT 03 to Plan Our 2025 Marketing Campaign.
0:16:05 It was full of like actual insights as well as just things I had not thought of about how
0:16:06 to apply AI to marketing.
0:16:08 I highly suggest you check it out.
0:16:11 Listen to Marketing Against the Grain wherever you get your podcasts.
0:16:15 Cool.
0:16:18 So I heard you had some controversial takes on supplements.
0:16:20 You know, I take a few supplements.
0:16:22 My co-host, Matt Wolfe, he’s super into supplements.
0:16:24 I think he takes like, I don’t know, 30 or so.
0:16:27 But yeah, I heard you had some controversial takes on supplements.
0:16:29 So like, should I be taking supplements?
0:16:29 Do they matter?
0:16:32 Or are they like, you know, are they BS or what?
0:16:35 I think supplements are overrated and many are likely BS.
0:16:36 Okay.
0:16:39 And at the same time, I take 30 to 50 a day.
0:16:41 In case, okay.
0:16:42 Kind of in case, I don’t know.
0:16:44 I think I’m like updating my views on it.
0:16:45 So there’s a few things.
0:16:53 One is that no amount of supplementation is going to solve for foundational problems that
0:16:56 should be addressed by attacking the root cause.
0:17:01 And no amount of supplementation is going to reliably guarantee you live past 100.
0:17:01 Right.
0:17:02 Right.
0:17:06 So right away, we know that even if supplements are marginally useful, they’re not that game
0:17:06 changing.
0:17:11 So we can already say, if they’re not that game changing, then I shouldn’t obsess about
0:17:11 them too much.
0:17:17 The next thing I’d say is that a lot of supplements are fake or dosed incorrectly or have other
0:17:21 sorts of additives or use the wrong part of the compound.
0:17:25 I was in the UAE last week and I was going to buy some vitamin D.
0:17:27 There are all these things in the pharmacy.
0:17:28 And you see this in the US as well.
0:17:31 There are all of these vitamin D in the pharmacy with like fancy branding.
0:17:33 I’m like, okay, but let’s actually look into it.
0:17:33 I look into it.
0:17:35 It’s like 300 I use.
0:17:38 And I’m like, well, no, I need like at least probably 5,000 for what I’m going for.
0:17:41 So 300, 5,000, that dosing doesn’t seem right.
0:17:46 The actual supplement itself has preservatives and colors and fake sugar and other shit in
0:17:46 it.
0:17:47 I’m like, well, I don’t want that.
0:17:49 Is that because it’s cheaper to make or something like that?
0:17:49 Is that why they do that?
0:17:50 I don’t know.
0:17:54 And then it didn’t have vitamin K, which is necessary for the absorption of vitamin D.
0:17:58 So it’s like, great, the marketing’s good, but the vitamin D is a negligible amount.
0:18:03 I’m adding all of this other shit to my body and there’s no vitamin K, which is a co-factor
0:18:04 necessary for the absorption of vitamin D.
0:18:11 Now take a correctly dosed vitamin D, really good source, purely encapsulated, no additives,
0:18:14 3,000 I use, maybe 5,000 I use.
0:18:16 They add vitamin K2.
0:18:20 Ideally, they add vitamin K2 as both forms, MK4 and MK7.
0:18:22 And maybe they add vitamin K1 as well.
0:18:26 All of those things now make a form of, that’s just vitamin D.
0:18:29 You can look across every other supplement and you have a similar thing.
0:18:32 So the second thing I’d say is like, not all supplements are created equal and there’s
0:18:33 a lot of junk out there.
0:18:37 And then the third thing I’d say, which is more speculative, and I don’t believe this
0:18:41 enough yet to stop taking supplements, but there might be something here, is that supplements
0:18:42 actually drain energy.
0:18:45 Our bodies are not used to processing the dry powder.
0:18:50 A lot of negative charge goes into actually creating the supplements in a big manufacturing.
0:18:50 lab.
0:18:55 And as a result, through mechanisms we don’t quite understand, supplements drain your body
0:18:56 of energy.
0:18:59 And again, I don’t believe that enough to have stopped taking them.
0:19:00 I took 30 last night.
0:19:02 But I think it’s interesting.
0:19:02 Right.
0:19:05 And some people hear this stuff and they’re like, no, that’s bullsh**.
0:19:09 Typically, when I hear someone say that’s bullsh**, my response is, this is interesting.
0:19:13 We don’t have evidence yet, but it might be worth pursuing and studying.
0:19:15 Because before there’s evidence, there’s always no evidence.
0:19:16 Yeah, totally.
0:19:16 Right?
0:19:17 It’s just a hypothesis.
0:19:22 Probably for the last two years, in terms of longevity, everyone thinks of Brian Johnson
0:19:22 now.
0:19:26 What do you think of Brian and what he’s doing and trying to live forever and all the crazy
0:19:27 stuff he’s doing on his body?
0:19:29 Look, I have a lot of respects for Brian.
0:19:30 Yeah.
0:19:32 I think that he’s a very good thinker.
0:19:35 I think that he’s creating a religion, which is non-trivial.
0:19:38 I think he cares about what he’s creating as a religion.
0:19:45 There are a set of very strong fundamental beliefs underlying it, which is that we need to follow
0:19:48 the algorithm because there’ll come a point in time where the algorithm knows more than
0:19:53 us, where the only thing we know is that at this very moment, I don’t want to die and
0:19:57 therefore don’t die is the most played game in a moniker or the religion.
0:19:57 Right.
0:19:58 So I really like that.
0:20:01 I think at the same time, he’s popularizing several health interventions.
0:20:03 I don’t think he’s at the frontier.
0:20:07 I see things he does and I’ll have doctor friends say, this is dumb.
0:20:08 He’s going to change this.
0:20:10 And then like three months later, he like changes it.
0:20:10 That’s fine.
0:20:11 He’s learning.
0:20:14 You got to remember, he’s been playing this game for five years.
0:20:17 There are doctors out there who have been playing this game of human optimization for
0:20:20 like 30 years, 40 years, 50 years.
0:20:24 So I still think there’s a lot he doesn’t know and he’s discovering it and sharing everything
0:20:24 he’s learning on the way.
0:20:26 So a lot of respect for Brian.
0:20:26 Right.
0:20:28 Yeah, I think that’s part that’s interesting.
0:20:32 And, you know, what I found fascinating in like a few interviews with him, he talked about
0:20:35 one of the reasons for him doing it was that, you know, it’s kind of like the whole thing
0:20:39 I was saying earlier, like we’re heading into the age of AI, like some of this technology
0:20:42 may unlock us to live an extra hundred years in the future.
0:20:47 And so, hey, try to make sure you can make it to that point so you can actually extend
0:20:47 your life.
0:20:49 I think that’s an interesting point.
0:20:49 Yeah.
0:20:52 And one thing I was thinking about, people always talk about their sleep, like monitoring
0:20:53 their sleep.
0:20:54 And I think you mentioned you used to do that.
0:20:56 Do you recommend monitoring your sleep?
0:20:59 Because that’s something I’ve always thought I probably should be doing that.
0:21:01 But I’m worried that I’m going to get super obsessed with it.
0:21:04 And maybe that’ll make me sleep, you know, less actually, because I’m worried about it.
0:21:07 I think most people should try and see whether they like it.
0:21:11 I think most people should do it for at least a couple of months because they’ll learn,
0:21:16 they’ll start to be able to develop a better interoception such that when they remove the ring
0:21:20 or the band, they still have a sense of what’s going on with their sleep.
0:21:25 I do think that it can end up making you more anxious about sleep and that will make your
0:21:26 sleep worse.
0:21:29 And that’s part of why I don’t open my Aura Ring app every morning now.
0:21:34 But I also think that even if you don’t want to look at the data, it could be worth collecting
0:21:37 the data because we’re into a world where we never have to look at the data.
0:21:39 The AI will just tell us what to do.
0:21:43 And in that world, I want to have a lot of data on myself to support the AI.
0:21:45 So that’s kind of how I think about the sleep question.
0:21:47 With Superpower, do you have something like that for now?
0:21:50 Do you like submit sleep data or is that something you guys have thought of doing?
0:21:53 Yeah, we integrate with Aura, several other wearables.
0:21:54 Yeah.
0:21:56 And that allows us to collect sleep data.
0:21:57 Awesome.
0:22:01 So you mentioned, you know, you already kind of shared like one controversial health belief
0:22:03 about the thing with supplements, they might drain energy.
0:22:06 Is that the most controversial health belief that you have?
0:22:08 Or is there anything else that’s even more controversial?
0:22:13 There are a lot of things that I think have a 5% probability, which most people call like
0:22:14 bulls**t.
0:22:16 And there are some that I think are higher probability.
0:22:17 Okay.
0:22:20 Now, my problem with all the 5% probability ones is I’ll say them and people say, this is
0:22:21 what he believes.
0:22:22 I’m like, no, no, no.
0:22:22 I believe the opposite.
0:22:24 But I think there’s a chance of this thing.
0:22:25 People have a hard time.
0:22:29 They’re like, have really binary mindsets of like, no, it’s got to be 100% true.
0:22:31 Or he’s saying it’s 100% false.
0:22:33 And people have a hard time in between.
0:22:33 Yeah.
0:22:33 Totally.
0:22:37 I guess like a classic example of this is vaccines.
0:22:43 I do think there’s somewhere between a 5% and probably 15% chance the vaccines are linked
0:22:45 to autoimmunity and maybe even autism.
0:22:46 There’s a low chance.
0:22:47 There’s a low but real chance.
0:22:51 I will say something like that and people will be like, he’s an anti-vax.
0:22:52 I’m like, no, I said there’s a low chance.
0:22:57 And therefore, we have to explore it and consider it versus just criticize and turn the blind
0:23:00 eye and hate and label, right?
0:23:00 Yeah.
0:23:03 Hating and labeling sounds a lot like propaganda to me.
0:23:06 And there are certainly incentives to obstruction industry that way.
0:23:12 One belief that I am higher probability on is one of our advisors and doctors, she runs
0:23:14 one of the leading cancer clinics around the world.
0:23:18 And I asked her, what is the number one thing you can do to prevent cancer?
0:23:22 And she says, PMA, positive mental attitude.
0:23:23 It’s the number one thing.
0:23:24 And I actually agree with that.
0:23:30 I genuinely believe that how we think, the beliefs we have, the self-talk, the way we relate
0:23:35 to ourselves, the amount of gratitude we have, impacts all health outcomes, impacts cancer,
0:23:39 neurodegeneration, autoimmunity, pain, and other things in our body.
0:23:44 And I think that the relationship between the brain and thoughts and biology is not well
0:23:44 understood.
0:23:47 And we’ll increasingly get to a world where it is well understood.
0:23:50 Do you have any theories on if that’s true?
0:23:51 Like why it’s true?
0:23:55 Like why would positive thinking and things like that have a positive impact on your health?
0:23:57 Besides maybe like it lowers your stress or there’s something else more extreme?
0:24:01 Like, look, a lot of our biology is modified by our brain.
0:24:02 We know that.
0:24:06 Like we can, through our brain, choose to increase or decrease our heart rate.
0:24:07 Right.
0:24:07 Right.
0:24:09 We can choose to do a lot of things.
0:24:10 We can tense our muscles.
0:24:13 We can do a lot of things via our brain.
0:24:19 And we understand how thinking certain things changes biology and changing a biology changes.
0:24:20 We understand that mapping.
0:24:24 Now, what we don’t understand is the mapping when it gets to things like cancer.
0:24:29 But I think if the mapping exists for very simple things, I think that it’s likely and
0:24:32 completely possible that it exists for more complex things as well.
0:24:36 The exact mechanism I don’t understand, I don’t know if anyone understands, which is
0:24:38 why people still call this woo-woo and pseudoscience.
0:24:43 But throughout history, many of the things we now call science, we used to call pseudoscience.
0:24:45 So again, I think the right approach is curiosity.
0:24:50 And that’s the first step in the scientific method, curiosity and hypothesis, rather than just
0:24:50 being closed-minded.
0:24:51 Yeah, I agree.
0:24:54 In some ways, you know, science has become almost like a cult.
0:24:55 You know, everyone has to agree.
0:24:56 There’s, you know, what’s the paper?
0:25:00 And if it’s not a paper that their peers have reviewed, you can’t even discuss it.
0:25:02 And it’s like, no, you can still discuss it.
0:25:06 Like, it doesn’t mean it’s 100% true, but it doesn’t mean you should just like completely
0:25:08 shut the door and say it’s impossible.
0:25:09 That’s always kind of driven me crazy.
0:25:10 Okay.
0:25:13 So maybe I got a fun question to end things.
0:25:16 You know, imagine you have a time machine, you know, and you step out.
0:25:17 It’s 2050.
0:25:18 What’s different?
0:25:19 What’s changed?
0:25:21 You know, not just health, but maybe also talk about health.
0:25:24 Gosh, no one knows.
0:25:30 More than ever, no one knows because the rate of change is completely unprecedented.
0:25:35 I think that it’s going to be way different to what people think.
0:25:39 I suspect that we likely won’t think about what we do in our actions.
0:25:43 I think the AI will just tell us and we will blindly follow the AI.
0:25:49 I think that just about everything we today call work will no longer exist as jobs.
0:25:52 So you’re most thinking like, I mean, we almost become the robots.
0:25:53 Like the AI is powering us.
0:25:54 Well, potentially.
0:25:55 Yeah.
0:26:00 I think that in a world where the AI knows far, far, far more than us, we don’t have a choice
0:26:01 but to follow the algorithm.
0:26:05 So there’s totally a world where the AI determines our behaviors.
0:26:10 There’s also a world where the AI has merged with us and the AI is part of our thoughts.
0:26:15 And we are cyborgs and I think that cyborgs will be here far closer than we think.
0:26:21 Particularly because if someone who’s a cyborg has a survival advantage, then other humans
0:26:24 will have to become cyborgs because we’re reasonably Darwinistic creatures.
0:26:26 We care about survival advantages.
0:26:29 And as a result, I think cyborgs are here sooner than people think.
0:26:33 Like by 2050, are we like merging with AI by then?
0:26:35 Yeah, I think so.
0:26:37 Yeah, if I was doing an overrun.
0:26:37 Yeah, I think so.
0:26:40 Somebody’s listening right now like, what the hell are they talking about?
0:26:42 Yeah, I know.
0:26:42 Cool.
0:26:43 Oh, yeah.
0:26:44 One last question.
0:26:46 I always like to ask people, you know, my son is 11.
0:26:50 And, you know, now with AI, I’m always trying to think like, what should I be teaching him,
0:26:52 you know, to make sure he’s successful in the future?
0:26:54 If you had a child, what would you be teaching them right now?
0:26:55 Would you be teaching them to code?
0:26:58 Or would you, you know, something completely different?
0:27:01 No, no, no skills, no skills, no jobs, none of it’s relevant.
0:27:09 I would be teaching them people and relationships and how the world works and leverage and company
0:27:10 building.
0:27:13 Maybe I’d get them to like play around with AI and understanding it.
0:27:15 But like the AI is going to do the coding.
0:27:17 The AI is going to do the prompting.
0:27:19 I would focus very much on people.
0:27:23 I think that scarce resources will not be, can you code?
0:27:26 It’ll be, can you unlock a door that is gated by a person?
0:27:27 Right.
0:27:31 So I’d focus more on everything that’s innately human, rather than everything that is some
0:27:35 sort of skill that is constructed by the current nature of our world.
0:27:37 That makes a lot of sense.
0:27:37 Yeah.
0:27:41 So like teaching him more about culture, comedy, just all kinds of different things that would
0:27:43 be useful in personal relations, probably.
0:27:51 Psychology, biology, getting them to meet people, host dinners, sales, persuasion.
0:27:52 Interesting.
0:27:56 Like, I think the importance of networks likely become far greater.
0:27:59 This whole idea of like, it’s not what you know, it’s who you know.
0:28:04 I think that that is more true, likely in the next five years than it has been for the
0:28:06 past couple of decades.
0:28:06 I agree.
0:28:09 You know, I’ve had this conversation with my friend, Greg Eisenberg.
0:28:11 You know, it’s part of the reason we both have been doing more content.
0:28:13 We totally agree with it.
0:28:17 Like in the future, your networks and who you know is going to matter a whole lot, especially
0:28:19 when anyone can press a button and copy a product.
0:28:20 It’s going to matter a lot.
0:28:21 I think so.
0:28:26 But I also wonder at the same time, whether we’re going to just see AIs chatting with AIs.
0:28:32 For example, right now, if someone messages me on LinkedIn, my EA is responding most of
0:28:33 the time rather than me.
0:28:37 Now, in a year, that’s probably going to be my AI responding.
0:28:39 Is this the first time we’ve actually talked?
0:28:42 No.
0:28:45 That’d be hilarious.
0:28:46 I still got Twitter.
0:28:46 Twitter’s less noisy.
0:28:49 LinkedIn, I get like hundreds of messages a week.
0:28:51 Twitter’s still quite high signal.
0:28:52 So I’m on Twitter.
0:28:53 I’m using WhatsApp.
0:28:55 Emails are a mix.
0:28:55 Half, half.
0:29:00 And again, though, we’ll get to a world where the AI drafts our emails, where the AI liaises
0:29:01 with the other AI.
0:29:03 It’s just like AIs all the way down.
0:29:06 I think that the problem with the world today is we treat kids like they’re kids.
0:29:08 And we never used to do that.
0:29:11 Alexander the Great was 21 when he ran the army.
0:29:12 Napoleon Bonaparte was 22.
0:29:14 Julius Caesar was 23.
0:29:16 Joan of Arc was like 20.
0:29:19 I might be getting the ages off by a couple of years, but roughly that.
0:29:22 And that’s because when they were kids, they were not treated like kids.
0:29:27 And I think that gaming when we’re younger allows us to do something like this.
0:29:29 And then you can start creating your own businesses.
0:29:32 I kind of reflect back on myself when I was young.
0:29:37 I was like, I actually think I was better as a business person on Minecraft than I was
0:29:42 for the next five to 10 years in real life because I had zero, zero fear and no one told
0:29:43 me whether I could or couldn’t do something.
0:29:44 Yeah, totally.
0:29:48 There were so many strategies that worked to make money on Minecraft that honestly would
0:29:51 have worked really, really damn well in the real world.
0:29:51 Right.
0:29:53 I just didn’t realize I could do it in the real world.
0:29:54 And I think about that often.
0:29:55 Right.
0:29:56 Dude, this has been awesome, Max.
0:29:58 It’s been great getting to know you.
0:29:58 Yeah.
0:29:59 Likewise, Nathan.
0:30:01 Yeah, it’d be awesome to have you on again sometime in the future.
0:30:02 Thanks, Max.
0:30:02 Great.
0:30:03 Thanks, Nathan.