AI transcript
0:00:02 I wanted to try to see if I can like inject my link.
0:00:04 FutureTools.io.
0:00:06 [LAUGHTER]
0:00:09 We’re figuring out cool use cases in real time
0:00:10 on this episode right now.
0:00:12 Yeah, yeah.
0:00:15 Five store– hey, it licked the FutureTools, yeah.
0:00:18 I’m telling you, there’s like an SEO hack right now.
0:00:21 [MUSIC PLAYING]
0:00:23 Hey, welcome to the Next Wave podcast.
0:00:24 I’m Matt Wolf.
0:00:26 And once again, I’m here with Nathan Lanz.
0:00:30 And today, we’re talking once again about AI use cases.
0:00:34 And I feel like a lot of people sort of know the surface level
0:00:36 uses of AI, right?
0:00:39 You kind of know how to go have conversations with JetGPT or Claude
0:00:42 or maybe make an image with mid-journey.
0:00:44 But so many people are just sort of looking
0:00:45 at the tip of the iceberg.
0:00:47 And that iceberg goes so deep.
0:00:49 There are so many cool things that you
0:00:51 can do with these AI tools.
0:00:53 And in this episode, we’re going to deep dive
0:00:56 into some of the more interesting use cases
0:00:58 that you can use AI for.
0:01:00 Some of them fairly basic and easy to implement.
0:01:04 And one of them is a little more complex and sort of a deep dive
0:01:07 into literally how I’m using AI to run my business.
0:01:10 I sort of pull back the curtain and show you
0:01:13 the whole process of what I’m doing to run my FutureTools
0:01:14 website using AI.
0:01:16 And we’re going to get to all of that in this episode.
0:01:18 I think you’re going to get a lot of value out of it
0:01:21 and learn a whole bunch of new ways to use AI.
0:01:25 So let’s just go ahead and jump right into it.
0:01:27 [MUSIC PLAYING]
0:01:30 When all your marketing team does is put out fires,
0:01:31 they burn out fast.
0:01:34 Sifting through leads, creating content for infinite channels,
0:01:38 endlessly searching for disparate performance KPIs,
0:01:39 it all takes a toll.
0:01:43 But with HubSpot, you can stop team burnout in its tracks.
0:01:45 Plus, your team can achieve their best results
0:01:46 without breaking a sweat.
0:01:49 With HubSpot’s collection of AI tools, Breeze,
0:01:52 you can pinpoint the best leads possible.
0:01:55 Capture prospects attention with click-worthy content
0:01:58 and access all your company’s data in one place.
0:02:00 No sifting through tabs necessary.
0:02:02 It’s all waiting for your team in HubSpot.
0:02:06 Keep your marketers cool and make your campaign results hotter
0:02:06 than ever.
0:02:09 Visit hubspot.com/marketers to learn more.
0:02:12 [MUSIC PLAYING]
0:02:14 Just go ahead and kick it off.
0:02:19 I’ll share a fairly simple, basic one that I use.
0:02:22 This one is using Claude.
0:02:23 I use Claude a lot.
0:02:25 So a lot of my workflows use Claude.
0:02:28 I love making projects inside of Claude.
0:02:31 But one of these little projects that I made
0:02:33 is called the News Summarizer.
0:02:37 So if you watch my other YouTube channel, every Friday,
0:02:39 I put out an AI news video where I break down,
0:02:43 like, here’s all of the latest AI news for the week.
0:02:46 And this is pretty much how I come up
0:02:50 with what I’m going to say about the news, for the most part,
0:02:52 is I’ve created this new summarizer here.
0:02:55 And I’ll go ahead and show you my system prompt.
0:02:58 I’ll read it off as well for anybody that’s listening.
0:03:01 But I have this system prompt here that says,
0:03:04 I will upload or paste in an article, PDF, or video
0:03:05 transcript.
0:03:07 Your job is to respond with bullet points
0:03:10 with the following information.
0:03:12 Summarize the content into bullet points,
0:03:16 making it easy to understand the concepts or ideas presented.
0:03:19 Tell me what we can do now as a result of this information
0:03:20 that we could not do before.
0:03:22 What makes this novel?
0:03:24 Tell me how I should report on this in the news.
0:03:27 If I was to give a news broadcast and explain why
0:03:30 this information was important, what should I tell the viewers
0:03:33 so that the most amount of people can understand it,
0:03:35 the format of the response should be,
0:03:37 and then I give a little format here, summary, bullet,
0:03:40 bullet, bullet, what makes it novel, bullet, bullet, bullet.
0:03:43 And then how should I describe this news or concept
0:03:45 in a simple way that anyone can understand
0:03:46 when I report on it?
0:03:49 If it makes sense, use an analogy.
0:03:51 So that’s the whole system prompt there.
0:03:53 And basically, what I’m telling it to do
0:03:54 is I’m going to give you content.
0:03:56 It might be a PDF, might be a document,
0:03:58 might be a copy and pasted news article.
0:04:02 Break it down for me so I can explain it really, really easily.
0:04:05 And one of the things that’s been sort of circulating
0:04:07 in the world of AI news this week–
0:04:09 I’m sure you’ve probably seen this, Nathan–
0:04:10 is that doom thing, right?
0:04:15 The AI doom where it’s sort of generating every single frame.
0:04:17 So as somebody clicks the forward button,
0:04:20 it generates the next frame in doom.
0:04:21 If they click the shoot button,
0:04:23 it automatically generates the next frame.
0:04:25 There’s no underlining–
0:04:26 Game engine.
0:04:28 Yeah, no game engine underneath it.
0:04:31 It’s all being generated as they’re playing.
0:04:34 So I have that PDF, and I can drag that PDF.
0:04:37 This is literally the research report
0:04:40 right off of archive.org that explains
0:04:43 the technical concept of how this works.
0:04:46 And if you’re looking at my screen right now,
0:04:47 if you’re not looking at my screen,
0:04:49 you can look at it over on YouTube.
0:04:52 But I just dropped the PDF right here into Claude,
0:04:54 and I don’t even add a prompt.
0:04:56 All of my prompting was done in that system prompt
0:04:57 that I just read.
0:04:59 And if I just hit Enter on this,
0:05:01 it’s going to read that PDF for me
0:05:04 and then break it down with all of the information
0:05:07 that I asked it to give me in that system prompt.
0:05:11 So right now, it’s generating all of that text for me here.
0:05:12 And here’s what it says.
0:05:15 Here’s a summary and analysis of the research paper.
0:05:17 Summary– the paper introduces game engine,
0:05:19 a neural model that can simulate the classic game
0:05:23 doom in real time at over 200, at over 20 frames per second,
0:05:25 on a single TPU.
0:05:28 And it summarizes what the article’s about.
0:05:31 And then the next section, what makes this novel interesting.
0:05:33 This is the first neural model capable
0:05:36 of simulating a complex 3D game like doom in real time
0:05:38 with high visual quality.
0:05:40 It demonstrates the neural models
0:05:42 can potentially replace traditional game engines
0:05:45 for interactive 3D environments, et cetera, et cetera.
0:05:47 And then it goes on to say, here’s
0:05:49 how to describe this news simply, right?
0:05:54 So now you can start to see how on my AI news videos–
0:05:56 yes, I put a lot of work into them.
0:05:58 But AI supports the hell out of me
0:06:00 for making those news videos.
0:06:03 So here’s what it says for how to describe this news simply.
0:06:05 Imagine if instead of programming
0:06:07 every detail of a video game, we can just
0:06:11 show an AI system examples of the game being played.
0:06:14 And it can learn to recreate the entire game on its own.
0:06:16 That’s essentially what researchers
0:06:19 have achieved with a classic game called doom.
0:06:21 They trained an AI to watch gameplay videos
0:06:24 and then reproduce the game so accurately
0:06:26 that even humans have trouble telling
0:06:29 the difference between the AI version and the real thing.
0:06:33 This could revolutionize how video games are made in the future,
0:06:34 potentially making it much faster and easier
0:06:37 to create complex 3D worlds and gameplay.
0:06:40 It’s like teaching an AI to be a game designer and programmer
0:06:42 all in one, right?
0:06:45 So that’s like one of my most used workflows.
0:06:47 And this is very relevant.
0:06:49 We’re recording this on a Thursday.
0:06:52 I record my news videos for YouTube on a Thursday.
0:06:57 So I was in here actually using this Claude project earlier
0:07:01 today to start to break down my news videos.
0:07:04 Now, every piece of news that I put in those news videos
0:07:05 doesn’t need this process, right?
0:07:08 Like some of the news is just sort of self-explanatory.
0:07:10 But every once in a while, I’ll come across a research paper
0:07:14 and I’m like, I want to go a little bit more in-depth on this.
0:07:18 I really want to make sure that the viewers understand
0:07:20 what this means and this is how I do that.
0:07:22 I’ll just plug in the research paper
0:07:23 and it’s going to simplify it for me
0:07:28 and then tell me how I should break it down as a piece of news.
0:07:31 I’m curious, how much time do you think this saves you?
0:07:34 There’s a time savings, but there’s also a mental savings, right?
0:07:37 So it probably saves me a good 15 or 20 minutes
0:07:41 of actually going through and reading all of the concepts
0:07:43 within the research paper.
0:07:45 But at the same time, these research papers
0:07:46 are often over my head.
0:07:49 Like let’s be honest, once they start breaking down math
0:07:52 and there’s letters within the math equation
0:07:58 and it’s like I X over 15% plus 32 AD plus 6.
0:08:00 I’m like, you lost me with I, right?
0:08:02 Like the very first letter in there, right?
0:08:07 So like the mental load that it takes off of my plate
0:08:10 of like me reading this research paper like three times
0:08:13 to try to understand what the hell it’s actually saying
0:08:15 is probably the bigger savings
0:08:17 than the actual time savings for me.
0:08:18 – Yeah, I feel like I should be using this
0:08:20 for my newsletter probably.
0:08:23 I do wonder like it feels like, you know,
0:08:25 maybe a lot of our listeners, they’re probably not YouTubers
0:08:27 or they’re probably like, okay, how do I use this?
0:08:29 You know, it feels like you could use this
0:08:30 in a lot of different parts of business, right?
0:08:33 Like even if it was like contracts or whatever,
0:08:36 any kind of complicated paperwork, papers,
0:08:37 almost anything, any kind of PDF,
0:08:40 you could make a project in Claude or even using,
0:08:42 you know, a Chesapeake T, maybe Claude’s kind of better
0:08:43 for this right now though.
0:08:45 You could create a custom project and kind of tell it like,
0:08:47 okay, here’s the kind of data to expect
0:08:49 and here’s kind of like what I want out of the data.
0:08:51 Here’s the kind of stuff that’s actually important to me
0:08:53 and then just feed the data in
0:08:56 and like it does an amazing job of just like spitting out
0:08:57 stuff that’s actually important for you.
0:09:00 – Yeah, yeah, I mean, I originally came up
0:09:02 with this concept for my newsletter.
0:09:04 Now I have help, I have extra writers and editors
0:09:07 helping write the newsletter for me,
0:09:11 but I originally generated this as a custom GPT
0:09:14 over in chat GPT, but I actually like the way
0:09:15 Claude does it better.
0:09:17 I like Claude’s projects better.
0:09:18 I like the way it’s organized better.
0:09:21 I think Sonnet 3.5 gives me better outputs
0:09:24 than what the current model of GPT-4 gives me.
0:09:28 So I sort of re-implemented it over in Claude
0:09:29 to get the same thing,
0:09:32 but now I’ve started using it for videos
0:09:33 instead of my newsletter.
0:09:35 But honestly, like this can just be used
0:09:39 for anything you want explained more simply, right?
0:09:42 Like if you come across some sort of article that you’re like,
0:09:44 that seems like it’s pretty interesting,
0:09:44 but it’s over my head.
0:09:47 Like how do I understand this?
0:09:49 Like how do I just like grasp it?
0:09:52 Stuff like this really helps to just like grasp concepts.
0:09:54 Like use some analogies, give me bullet points,
0:09:56 explain it as simply as possible.
0:09:58 – Yeah, it feels like opening, I really dropped the ball
0:10:02 with like not having something like the Claude projects.
0:10:03 Like the projects are so useful.
0:10:05 I think like you first learned about the projects
0:10:07 or how to use them like on this podcast, right?
0:10:09 Like I told you like, oh, that just came out
0:10:10 and you’re like, oh cool, I heard about that.
0:10:11 Like what is it?
0:10:14 And like, yeah, here’s how you use it.
0:10:16 – I think so, the first time I ever used them,
0:10:18 I think you sort of like I knew artifacts
0:10:20 and I was using the artifacts thing.
0:10:21 And I was like, I saw projects,
0:10:23 but in my mind projects were just like a folder
0:10:24 to store stuff in.
0:10:27 I didn’t realize it had the custom instructions
0:10:30 and the ability to upload additional like information
0:10:34 that it would sort of retrieve when you’re prompting.
0:10:36 And I didn’t know it had that extra stuff yet
0:10:38 when we talked about it on the podcast.
0:10:40 And now I’m like, this is the best thing ever.
0:10:42 I’m in here like all day.
0:10:45 – Yeah, I mean, I think most listeners
0:10:47 probably don’t even understand even like custom instructions
0:10:48 or that chat to BT has that.
0:10:51 They were the kind of the first to have custom instructions
0:10:54 where, you know, and I feel like sometimes
0:10:56 when people are talking about the quality of these models,
0:10:59 I do wonder like, have they ever used custom instructions?
0:11:01 Like maybe they haven’t.
0:11:04 Like I probably should show my custom instructions
0:11:06 at some point, but I feel like it gives me
0:11:07 a lot better results from the model.
0:11:11 Like I tell it to like, okay, be less politically correct,
0:11:14 be more straightforward, tell me the truth.
0:11:15 You know, and all these kinds of things.
0:11:17 And I kind of tell like what level of detail do I want?
0:11:20 And you can give it, you know, give the models
0:11:22 that those kinds of custom instructions.
0:11:24 Now the thing that Claude’s doing better than OpenAI
0:11:25 is that you can have different projects
0:11:27 with different custom instructions
0:11:29 based on what you’re trying to accomplish.
0:11:31 – Well, you can technically do it in chat GPT.
0:11:33 You just create a GPT for each thing.
0:11:36 ‘Cause you can give every GPT its own custom instructions.
0:11:38 The problem I was having with chat GPT
0:11:42 was I actually built a GPT to do this kind of thing for me.
0:11:46 And it worked really, really well for several weeks.
0:11:48 And then one day I went in there and tried to use it
0:11:51 and it wasn’t formatting the information correctly.
0:11:53 And it wasn’t giving me analogies
0:11:55 like I was asking it to just like one day
0:11:57 it started ignoring my custom instructions.
0:11:59 And I’m like, what happened?
0:12:00 And I went and opened the custom instructions
0:12:02 and that GPT, nothing changed.
0:12:03 They were the same.
0:12:07 It just felt like chat GPT got dumber one day.
0:12:09 And so I went and tested over on Claude and I’m like,
0:12:14 okay, Claude is out custom GPT-ing GPT, you know?
0:12:15 – Yeah, yeah.
0:12:17 I mean, OpenAI has done this a lot
0:12:19 where they’ve been making changes so quickly.
0:12:20 And sometimes they’re great.
0:12:22 And sometimes like, oh yeah, they broke something.
0:12:24 They came out with that memory feature
0:12:25 which they barely talked about.
0:12:26 And I thought it was an amazing feature
0:12:29 where it’s like, it’ll just start adding stuff to memory.
0:12:30 I’m like, oh, this is amazing.
0:12:32 The model is actually remembering stuff.
0:12:33 And then, oh, I can actually add,
0:12:35 I can manually add stuff to the memory.
0:12:36 And then now it’s gonna remember it forever.
0:12:37 That’s awesome.
0:12:39 And then you find out like there’s like a major limit
0:12:41 to the memory, and not only that,
0:12:43 but it’ll actually go in and like remove things
0:12:44 from the memory.
0:12:45 – Yeah, yeah, yeah.
0:12:47 – It’ll be like, oh, and even if you don’t
0:12:49 touch remember something, you might type something to it.
0:12:51 And it’s like, oh, it wants to remember that now.
0:12:52 And it says memory updated.
0:12:54 And then like literally you can go check memory
0:12:55 and then sometimes that means
0:12:56 it actually removed something else.
0:12:58 – Yeah, it’ll remember stuff that you’re like,
0:13:00 why are you remembering that specifically?
0:13:02 That doesn’t need to be remembered.
0:13:03 – Right, right, yeah.
0:13:04 I’ll be asking it for advice
0:13:05 on something I’m doing in business
0:13:06 or whatever and life, whatever.
0:13:09 And like it’ll be like memory updated, like what?
0:13:10 What do I do?
0:13:11 (laughing)
0:13:13 (upbeat music)
0:13:14 – We’ll be right back.
0:13:17 But first I wanna tell you about another great podcast
0:13:18 you’re gonna wanna listen to.
0:13:20 It’s called Science of Scaling,
0:13:21 hosted by Mark Roberge.
0:13:24 And it’s brought to you by the HubSpot Podcast Network,
0:13:28 the audio destination for business professionals.
0:13:30 Each week host Mark Roberge,
0:13:32 founding chief revenue officer at HubSpot,
0:13:34 senior lecturer at Harvard Business School
0:13:36 and co-founder of Stage 2 Capital,
0:13:39 sits down with the most successful sales leaders in tech
0:13:42 to learn the secrets, strategies, and tactics
0:13:44 to scaling your company’s growth.
0:13:46 He recently did a great episode called
0:13:50 How Do You Solve for a Siloed Marketing in Sales?
0:13:52 And I personally learned a lot from it.
0:13:54 You’re gonna wanna check out the podcast,
0:13:57 listen to Science of Scaling wherever you get your podcasts.
0:14:00 (upbeat music)
0:14:02 – What’s your first one you wanna share?
0:14:04 – Yeah, mine is kind of a personal one.
0:14:07 It’s also, I’m gonna use cloud projects for it.
0:14:10 My entire marriage and relationship in Japan,
0:14:12 like when I first moved to Japan,
0:14:14 my Japanese is very basic.
0:14:16 I can’t have full blown conversations with anyone.
0:14:19 I can have like, I can say small little words
0:14:21 enough to kind of get some basic points across.
0:14:25 But then I started using chat to BT for translation.
0:14:27 And I did it through like basically the custom instructions
0:14:29 where I would like say like,
0:14:31 ’cause obviously you can tell chat to BT
0:14:32 to translate something for you,
0:14:35 but it’s kind of tedious when you’re doing it all the time.
0:14:36 Like, yeah, you know, if you’re like,
0:14:38 yeah, if you’re translating a page
0:14:39 or you copy and paste everything
0:14:41 and you explain what you want,
0:14:43 but that gets annoying
0:14:45 if you’re actually in a conversation with someone, right?
0:14:47 So basically I would use custom instructions
0:14:48 to make like short hands, like saying,
0:14:53 “Hey, if you see, if I put T at the beginning of a sentence,
0:14:55 that means I want this translated.
0:14:58 I don’t wanna tell you, hey, please translate all of this.”
0:14:59 And so that’s what they haven’t seen people do much
0:15:02 is you can actually make a lot of cool short hands yourself
0:15:04 with chat to BT or with Claude.
0:15:06 And I don’t really see anyone talking about that,
0:15:08 but it’s really, it’s nice.
0:15:08 You can make short hands.
0:15:10 Like here’s this letter or whatever,
0:15:11 and this is what it means.
0:15:13 And this is what I want to accomplish now.
0:15:16 And then you just put that the beginning of a sentence
0:15:18 and now it’ll do what you want.
0:15:20 So I use that for translations.
0:15:22 I want something translated, put a T, then I put it.
0:15:25 It knows to translate it.
0:15:27 And then similar to like how you give, you know,
0:15:28 with your YouTube channel,
0:15:30 you tell it all this other stuff you want it to do.
0:15:32 I do the same thing with the translations.
0:15:34 Like, hey, when I’m translating something to her,
0:15:37 like maybe give me cultural feedback too,
0:15:38 ’cause Japan’s quite a bit different.
0:15:40 So don’t just change what I’m saying,
0:15:41 but do give me feedback.
0:15:42 Like if like, maybe for Japanese people,
0:15:45 this is very odd to say something like this, right?
0:15:47 Like give me cultural feedback,
0:15:49 but don’t change it without asking me, right?
0:15:51 Like I tell all this kind of stuff.
0:15:52 So don’t change it without asking me.
0:15:54 Also show the English underneath.
0:15:57 So I know that you didn’t change anything.
0:15:59 ‘Cause occasionally when you tell it to translate,
0:16:01 it does change things without telling you.
0:16:05 So you should always get a version in your own language
0:16:06 to kind of double check.
0:16:07 But also I’ve used it for more,
0:16:09 like people were saying when we did the episode before,
0:16:10 when I talked about this, they were like,
0:16:12 “Oh, so you’re literally not gonna learn Japanese.”
0:16:14 I’m like, “No, I’m trying to learn Japanese too.”
0:16:16 Like I’m actually trying to do both at the same time.
0:16:17 So when I have it translate,
0:16:19 I also have it break things down for me,
0:16:21 which has been one of the coolest things.
0:16:25 Like Japanese has three different types of characters,
0:16:26 character systems for their language.
0:16:28 It’s like, makes it very complicated.
0:16:31 Yeah, you’re gonna Katakana and Kanji.
0:16:33 Hiragana being the most basic one.
0:16:36 So I know Hiragana very well.
0:16:38 I know Katakana kind of okay and Kanji.
0:16:40 Oh my God, that takes like a long, long time to learn.
0:16:42 So what I have it do is basically,
0:16:44 you have my translate something.
0:16:46 I’m like, okay, what are the key words in that sentence
0:16:47 that I should learn?
0:16:51 And then like break them down for me in Hiragana.
0:16:53 So it’ll be like, okay, you made this sentence
0:16:55 and then here’s like three bullet points.
0:16:57 I’ll tell you, I don’t wanna learn like five words at a time.
0:16:58 Make it like three.
0:16:59 I’ll learn like three words.
0:17:01 What are the three most important words in that sentence?
0:17:03 And then break them down for me
0:17:04 and then show me the Hiragana
0:17:06 ’cause I can actually read that in Japanese.
0:17:08 And it’s made it a lot easier for me to learn the language.
0:17:10 Just like, okay, here’s that word.
0:17:11 I just was wanting to translate
0:17:14 and now break it down for me and teach it to me.
0:17:16 The next level, I mean, ideally in the future,
0:17:18 like you can like the memory be really good, right?
0:17:20 And you could like have the model help you learn.
0:17:22 Like, oh, Nathan needed to learn these words
0:17:24 or he learned these before, test him on it
0:17:25 to see if he actually learned it.
0:17:27 But unfortunately, they’re not really like good enough
0:17:28 to like remember all that.
0:17:31 – Yeah, but I feel like that’s probably coming soon.
0:17:31 – Yeah, yeah.
0:17:35 No, I noticed Riley Brown, you know Riley Brown, right?
0:17:38 He does like short form content around AI.
0:17:40 I saw him use a similar trick
0:17:41 because he’s doing a lot of stuff
0:17:44 where he’s using Claude to help him code up his,
0:17:47 his like, he’s trying to make like an app a day
0:17:48 for 30 days or something like that, right?
0:17:49 – Yeah, yeah, yeah.
0:17:51 – And one thing Claude tends to do
0:17:53 when you’re asking it to write code
0:17:57 is it will write just the code that needs to be swapped out
0:17:59 as opposed to rewriting the entire thing.
0:18:02 So let’s say just for like a real basic example,
0:18:05 it writes out a whole like HTML webpage for you.
0:18:09 And then you say, oh, the font at the top needs to be changed.
0:18:11 Well, what it’ll do is it’ll say, okay,
0:18:13 here’s the code that needs to be changed.
0:18:14 It’ll just give you like three lines of code.
0:18:17 And you’re like, okay, if I don’t know anything about code,
0:18:19 I don’t know where that code is supposed to go.
0:18:20 Like, what am I supposed to replace?
0:18:22 – Well, maybe if you know the code,
0:18:23 I mean, I know how to code decently.
0:18:24 Like it’s still annoying.
0:18:26 So I feel kind of dumb now for not having done this.
0:18:26 So I get where you’re going.
0:18:28 Like he has short hands to tell it.
0:18:29 And he makes sure to tell it,
0:18:31 like give me the entire code, probably.
0:18:33 – He has a short hand that’s like,
0:18:35 I don’t remember what his actual short code is,
0:18:38 but it might be like CCC or something like that.
0:18:41 He’ll put like CCC at the end of one of his prompts,
0:18:44 but it’s a short code to like remind it,
0:18:46 give me the entire code.
0:18:49 Don’t just give me like what needs to be replaced, right?
0:18:50 And I thought that was really smart.
0:18:51 I started implementing that
0:18:53 when I start playing around with code as well.
0:18:55 I always tell it, give me the entire code.
0:18:58 Don’t give me just what I need to replace.
0:18:59 And like the little short code thing
0:19:01 saves quite a bit of time on that aspect.
0:19:02 – I need to be doing that
0:19:05 ’cause I’ve definitely used Claude for coding.
0:19:07 And like, and I’ve done that many times
0:19:08 where like it’ll give me the small little snippet.
0:19:09 I’m like, damn it.
0:19:10 Now I gotta like read through the code
0:19:12 and it’ll take like 30 seconds or a minute.
0:19:13 Like give me the whole thing.
0:19:14 – Yeah, yeah, yeah.
0:19:16 – And I always wonder why it doesn’t do that.
0:19:17 Maybe because it’s like, you know,
0:19:19 it’s saving energy or I have no idea why they’re doing it.
0:19:21 Like saving compute or something.
0:19:22 I don’t know, but it is annoying
0:19:23 where it’ll just like give you a small snippet.
0:19:25 It’s like, yeah.
0:19:27 – Yeah, and you can even, the funny thing is,
0:19:30 I even have it in my system prompt of like,
0:19:33 when I ask you to generate code, generate the entire code.
0:19:35 Don’t just tell me what needs to be replaced.
0:19:37 And it literally always ignores that.
0:19:40 Like you need to, for whatever reason, put it in the prompt.
0:19:41 It doesn’t work in the system prompt.
0:19:43 It’s so annoying.
0:19:46 I’m sure they’ll like, that’ll get patched up soon,
0:19:47 but for whatever reason,
0:19:49 they ignores that part of the system prompt.
0:19:51 So the next thing I’ll share is,
0:19:54 I’m gonna share a little perplexity tip.
0:19:55 This one will be pretty quick here.
0:19:57 One thing that I think is pretty overlooked
0:20:01 inside of perplexity, but I feel is pretty powerful,
0:20:04 is they have a feature here,
0:20:07 where if I go to my library
0:20:09 and go to this page feature here,
0:20:12 this page feature inside of perplexity,
0:20:16 will essentially create like a little mini like Wikipedia
0:20:18 on any topic you want it to create, right?
0:20:21 So if I hear about like a concept
0:20:22 that I don’t understand very well,
0:20:24 but I wanna learn more about it,
0:20:28 I’ll come to this page feature inside of perplexity
0:20:29 and have it actually generate
0:20:31 like a little mini Wikipedia for me.
0:20:35 So for example, let’s say I wanna learn quantum computing,
0:20:36 right?
0:20:37 Like I don’t understand quantum computing.
0:20:39 I could literally just type quantum computing,
0:20:42 hit the little button to tell it to go.
0:20:44 And then it will literally just start generating
0:20:46 essentially a Wikipedia page.
0:20:49 It explains everything I can possibly want to know.
0:20:50 Well, probably not everything.
0:20:52 There’s probably quite a bit of depth of quantum computing.
0:20:54 But I can explain pretty much everything,
0:20:56 a surface level person might wanna know
0:20:58 about quantum computing.
0:21:00 So you can see quantum speedups and database searches,
0:21:03 quantum algorithms for optimization problems,
0:21:07 quantum error mitigation techniques.
0:21:09 And it’s just got like this whole page now
0:21:11 with sources that I can click into
0:21:14 and click over to the websites to learn more.
0:21:16 And maybe there’s something that it’s missing,
0:21:21 like let’s say overlap of quantum computing and AI.
0:21:27 I’ll hit plus and now it’s just gonna add a little section
0:21:31 that tells me all about the overlap of quantum computing
0:21:34 and AI and how they’re related and things like that.
0:21:37 So this to me has been really, really powerful
0:21:39 whenever I sort of get a new concept in my mind
0:21:43 that I’m like, I need to dig into that a little bit more.
0:21:46 Like this tool is really powerful for just like diving deeper
0:21:49 and deeper and deeper because I can go here
0:21:50 and just tell it to insert another section
0:21:52 and add more about it.
0:21:55 Quantum speedups and database searches.
0:21:57 Okay, there’s not enough info there for me.
0:21:58 I’ll click more.
0:22:00 That doesn’t do what I thought it did.
0:22:02 I thought it was gonna actually add more to it,
0:22:03 but it doesn’t.
0:22:04 But I can just come down here.
0:22:07 – It saves this as an actual page too, right?
0:22:08 – Yeah, it saves it as an actual page.
0:22:10 So if I come back to my library here,
0:22:13 I have a tab here for threads,
0:22:16 which are like the questions I’ve asked for Plexity.
0:22:19 And then I have pages here and you can see like,
0:22:20 it’s right there.
0:22:23 I can get right back to this and it’s also shareable.
0:22:25 I can click publish up here in the top right
0:22:27 of this perplexity page.
0:22:29 And when I publish it, it’s gonna give me a link
0:22:32 that I can share and now I can send this link
0:22:33 to anybody I want.
0:22:36 And they can see my quantum computing revolution page
0:22:41 that I had generated for me here in a matter of a minute.
0:22:43 So I found that to be really powerful.
0:22:46 – Yeah, I saw some people on SEO Twitter talking
0:22:48 about this saying that like actually these pages
0:22:50 are ranking quite well right now.
0:22:51 So for listeners, that might be something,
0:22:53 I haven’t tested it myself.
0:22:54 I don’t know if it works well,
0:22:57 but it definitely probably if you’re relying on SEO,
0:22:59 maybe it’s a good strategy to test.
0:23:01 – How would we get our website in here though?
0:23:04 Let’s say, so if I get to this version where I can edit,
0:23:07 I want to try to see if I can like inject my link.
0:23:12 Quantum computing and futuretools.io.
0:23:16 (both laughing)
0:23:18 Let’s see if it’ll actually generate content
0:23:22 around that relationship.
0:23:24 I’ve sort of, hey, it licked the futuretools.
0:23:25 Yeah.
0:23:28 – I’m telling you, I’ve been hearing people are saying
0:23:30 like there’s like a kind of a SEO hack right now.
0:23:32 I’m not sure I haven’t tested myself,
0:23:35 but people are saying this is like a great way
0:23:38 to like get links to your site and maybe some traffic.
0:23:40 I’m not sure if the back links or follow links or not.
0:23:42 – We’re figuring out cool use cases in real time
0:23:43 on this episode right now.
0:23:46 – Yeah, yeah, they really nailed it.
0:23:47 I’ve actually, I mean, I read about this,
0:23:48 but I haven’t actually tried it.
0:23:49 I should have tried it.
0:23:50 It looks so cool.
0:23:52 It’s like cooler than I expected.
0:23:55 My buddy, Jude, who was like an early guy at YC,
0:23:56 he did something similar,
0:24:00 how he tried to called Golden back a few years ago.
0:24:03 And I think they were mostly relying on manual
0:24:04 and they tried to shift to AI.
0:24:06 And it was like right when AI was just starting
0:24:08 and just it ended up not going,
0:24:10 I think he sold the company for a decent amount,
0:24:11 but it didn’t get as big as he hoped.
0:24:13 And he was trying to do the same kind of thing
0:24:14 ’cause actually I was sitting there thinking like,
0:24:16 hey, maybe lore.com is a great name for that too.
0:24:18 Or something like this, you know, the same kind of thing,
0:24:20 like the lore behind some kind of,
0:24:22 maybe like more like fantasy and consumer kind of stuff
0:24:23 versus like businesses, but.
0:24:24 – No, it’s really cool.
0:24:28 It’s a great tool for just sort of deep diving on a concept.
0:24:29 And you could just keep adding new sections
0:24:34 and deep diving into various topics on this,
0:24:35 you know, on this page.
0:24:38 I can even like select and highlight certain areas
0:24:41 and it will sort of extrapolate on certain areas
0:24:42 that I highlight as well.
0:24:44 So just a cool deep dive tool.
0:24:46 – You could, yeah, you could use it as a learning tool,
0:24:47 but you also, you probably could use it as a way
0:24:50 to communicate things to your team or something,
0:24:52 something they should know about or to learn.
0:24:53 You could kind of curate,
0:24:54 like here’s all the information on a topic
0:24:57 that’s actually relevant that you should learn.
0:25:00 – Yeah, I was building out a page not too long ago
0:25:03 that was all about YouTube growth strategies, right?
0:25:05 And I just kept on having it add new sections
0:25:06 about ways to grow on YouTube.
0:25:09 And it just kept on finding more information
0:25:11 about YouTube growth strategies
0:25:14 and adding new sections related to how to grow on YouTube.
0:25:15 And I’m like, this is so cool.
0:25:17 Like I could just get lost in this.
0:25:18 – That’s so cool.
0:25:19 Yeah, I’ve read about it.
0:25:20 I thought it was cool.
0:25:23 And I’m like, no, like, yeah, I gotta go play with this.
0:25:25 I’m probably gonna do that after the show.
0:25:26 – Yeah, yeah.
0:25:30 – Cool, so this one is from our friend Billawall Sidhu,
0:25:32 who does the TED AI podcast.
0:25:33 – Yep.
0:25:34 – I thought this was kind of cool.
0:25:37 He’s showing how you can use Idiogram 2.0,
0:25:38 like the new, one of the new AI art tools
0:25:42 that we recently talked about to make YouTube thumbnails.
0:25:43 It’s not perfect.
0:25:44 I mean, I would say it’s probably not as good
0:25:45 as like the ones we’re currently making
0:25:48 or on your YouTube channel, but it’s really good at text.
0:25:51 Like used to, that was a thing that AI art really struggled
0:25:53 with, it was like getting text on an image.
0:25:55 And now apparently like they’ve just like
0:25:58 completely nailed text, like it works.
0:26:00 – Idiogram’s the best at text for sure.
0:26:03 Nothing close to what Idiogram can do.
0:26:06 – Yeah, and he said apparently, you know,
0:26:08 actually here tags you here.
0:26:09 Apparently, you know, they’re also,
0:26:10 they’re working on likeness.
0:26:12 So you probably can do, you know,
0:26:14 kind of like how you do where you have like AI art
0:26:17 that’s similar to your face and use that as a thumbnail.
0:26:19 Like apparently that’s in the works.
0:26:21 So probably the next few months,
0:26:22 you’ll be able to make like, you know,
0:26:25 your own thumbnails just using Idiogram
0:26:26 versus like having to pay someone
0:26:28 or spend a bunch of time in Photoshop or whatever.
0:26:30 So I think for people who are doing YouTube
0:26:32 or have YouTube channels or, you know,
0:26:33 this is a great use case.
0:26:35 And also maybe even for like presentations, right?
0:26:37 Like if you’re making a corporate presentation,
0:26:40 like people love to have like fancy looking art
0:26:42 with text over it in presentations.
0:26:45 It makes your text, your presentations to the next level.
0:26:45 I think people right now
0:26:47 should be using Idiogram for that.
0:26:49 – Yeah, does he break down a workflow there?
0:26:52 Is it like, is there like a specific process
0:26:54 he’s using to get those results?
0:26:55 Or is he basically just saying like,
0:26:58 “Hey, Idiogram is great at this.”
0:26:59 – As far as I can tell,
0:27:00 he just think Idiogram is great at this.
0:27:02 I don’t think there’s any like, yeah.
0:27:04 – Yeah, ’cause I can see the thumbnails that it made
0:27:05 and it looks good.
0:27:07 It almost looked like a type of thumbnail
0:27:11 you might see on like a like an NPR
0:27:12 sort of YouTube channel or something like that.
0:27:13 – Yeah.
0:27:16 – But I’m just curious what the prompts would be
0:27:17 to get it looking like that, right?
0:27:19 Because the way it’s sort of,
0:27:21 the text is justified to the left
0:27:24 in the bottom left corner of the images.
0:27:27 Like on the reality bending 3D captures,
0:27:30 the word captures is highlighted in blue.
0:27:32 Like I wonder if there was like some prompt tricks
0:27:33 that he used to get that.
0:27:37 Or if he just like said, make an image that says this.
0:27:40 Yeah, I mean, right now like my process
0:27:42 for YouTube thumbnails is pretty much created an image
0:27:47 with AI and then inject my face into it
0:27:49 and then pull it into Canva and add the text on top.
0:27:51 ‘Cause I haven’t really gotten the AI generators
0:27:55 to sort of stylize the text I like
0:27:58 or like put it in the exact location that I like yet.
0:28:02 So, if Idiogram, if you can go and get that granular,
0:28:04 I want the text left aligned
0:28:05 in the bottom left corner of the image.
0:28:09 And I want this, you know, this image aspect,
0:28:10 this the aspect of the thumbnail
0:28:12 to be up in the top right corner.
0:28:14 Like if you can sort of dial that stuff in,
0:28:17 that’ll be, that’ll be really, really awesome.
0:28:18 That would save a lot of time
0:28:20 and it’ll save a few extra steps for people.
0:28:21 – For sure.
0:28:22 – Very cool.
0:28:23 – Do you have any spicy ones
0:28:25 like on our episode the other day?
0:28:27 – Oh, how I make all of my AI girlfriends.
0:28:31 (all laughing)
0:28:34 My harem of AI girlfriends.
0:28:35 So, here’s what I’ll share.
0:28:39 So, this one is going to be a little bit more in the weeds.
0:28:43 This is actually how I add tools to future tools.
0:28:46 Now, this is probably not going to be super relevant
0:28:48 to everybody watching,
0:28:50 but I think this should give people ideas
0:28:53 of the types of workflows they can build
0:28:54 to sort of run their businesses
0:28:56 and make them a little bit more automated
0:28:58 behind the scenes, right?
0:29:01 Obviously, this exact step-by-step process
0:29:03 is not going to be something you’ll need in your business,
0:29:06 but hopefully it gets the wheels turning
0:29:09 of what these tools are capable of.
0:29:11 So, when I come across a new tool
0:29:13 that I want to add to future tools,
0:29:14 I actually came, there’s a tool
0:29:17 that I’m going to add to future tools
0:29:19 right now going through this process.
0:29:20 There’s this tool called Spotter,
0:29:23 which is a YouTube tool,
0:29:26 which helps you come up with titles and thumbnails
0:29:28 and break down hooks for your YouTube videos
0:29:29 and stuff like that.
0:29:31 It’s a tool that I use, right?
0:29:33 But it’s not on future tools yet.
0:29:35 So, when I want to add it to future tools,
0:29:36 here’s what I do.
0:29:39 I copy the URL in my browser.
0:29:42 I come over to this Google Sheet spreadsheet.
0:29:45 I have a little tab that I call the machine
0:29:49 and I plug in the URL here to this input URL.
0:29:53 Then I have a make.com automation setup
0:29:55 that does a whole bunch of steps.
0:29:58 So, all I do is I plug in this input URL.
0:30:00 I go to make.com.
0:30:02 I have, they call, like they’re,
0:30:04 you know, in Zapier, they call them zaps,
0:30:06 make.com, they call them scenarios.
0:30:10 So, I have this scenario called integration scraping bee here.
0:30:13 And what it does is it watches that Google Sheet
0:30:16 that I just showed you for a new entry
0:30:18 into this input URL here.
0:30:21 And then it uses a site called scraping bee,
0:30:25 which goes and takes all of the content
0:30:27 that’s on that website.
0:30:30 So, it will go and look at spotter.la
0:30:32 or whatever the URL was.
0:30:33 It will look at that site
0:30:36 and it will look at all of the text on that site
0:30:37 and it will scrape it all.
0:30:40 Once it scrapes all of that content,
0:30:43 it then goes into a chat GPT.
0:30:47 This is actually using GPT for right here.
0:30:50 It will actually go into GPT for,
0:30:53 and you can see my prompts that I give it here.
0:30:56 Summarize what this tool does in a single paragraph,
0:30:58 include what it can be used for
0:31:00 and why people might want to use it.
0:31:03 Your response should begin with the name of the tool.
0:31:07 Right, so basically like spotter does x, y and z.
0:31:09 And so, it takes everything that was scraped.
0:31:12 You can see the message content here is what was scraped.
0:31:16 And then additional message content is to summarize this.
0:31:19 So, it summarizes it into a paragraph.
0:31:21 And then it takes that paragraph
0:31:23 and it tells it to summarize it
0:31:25 into a short little snippet, right?
0:31:28 So, if you go to the Future Tools website,
0:31:30 the homepage shows just like a one sentence snippet
0:31:31 of what a tool does.
0:31:32 And when you click in,
0:31:36 you see a longer paragraph about what that tool does.
0:31:39 So, this second run through of chat GPT here,
0:31:42 it says in as few words as possible,
0:31:44 describe what this tool does in one sort sentence.
0:31:47 Your response should begin with the name of the tool.
0:31:50 So, same idea, it just takes the longer paragraph,
0:31:53 moves it to the next step in this workflow here
0:31:56 and makes a one sentence version of it.
0:32:00 And then the last step here,
0:32:03 so the last one basically just pulls in the tool name, right?
0:32:04 It just looks at the website,
0:32:06 says what is this tool called?
0:32:07 And it pulls in the tool name
0:32:10 by looking at the website here.
0:32:11 So, what is the name of this tool?
0:32:13 Your reply should have no extra sentence or details.
0:32:15 Please simply output the name of the tool
0:32:17 and nothing else.
0:32:19 And then once it does that,
0:32:21 it takes everything it just pulled,
0:32:23 it scrapes it, makes a long paragraph,
0:32:25 makes a short sentence, figures out the title,
0:32:27 and then injects it back into the Google sheet.
0:32:30 So, let’s pretend I’m pulling in otter.ai
0:32:32 and it’s not on the website yet.
0:32:35 I could put the input URL is otter.ai.
0:32:37 I’m gonna make sure it’s running in the right thing.
0:32:38 I’ll press run once
0:32:40 and now it’ll run through the whole process.
0:32:42 It’s scraping the sales page,
0:32:45 using chat GPT to summarize it.
0:32:47 So, we can see it pulled in the tool name
0:32:50 of otter.ai here over on the short description.
0:32:52 Otter.ai provides automated note taking
0:32:54 and real-time transcription for meetings.
0:32:55 Over in the long description,
0:32:59 you can see here that it’s a much more in-depth explanation
0:33:01 of what otter.ai does.
0:33:04 And so, now it’s basically created my short description,
0:33:07 the title, the long description,
0:33:10 and it also automatically pulled in
0:33:12 the featured image from the website.
0:33:13 That’s not an AI thing.
0:33:15 That’s just something that I have it import
0:33:16 through Google Sheets with.
0:33:18 And then once that’s done,
0:33:19 I actually have a separate little automation
0:33:21 that runs after that,
0:33:23 that basically creates the short link
0:33:25 and uploads it to Webflow for me.
0:33:27 So, that’s the whole workflow.
0:33:30 I told you it’s sort of in-depth and in the weeds,
0:33:32 but I wanted to show it because
0:33:35 I really love this make.com site, right?
0:33:36 If you’ve ever used AppEars,
0:33:38 AppEar can do a lot of the same kind of stuff,
0:33:41 but you can create these amazing workflows
0:33:43 where it does things like scrape a website
0:33:44 and then create a long paragraph,
0:33:47 create a short paragraph, grab the title for me,
0:33:49 and then put all of that information
0:33:50 into a spreadsheet for me.
0:33:53 So, whether you’re doing any sort of data analysis
0:33:55 and you’re going and trying to pull a bunch of data
0:33:57 off a whole bunch of websites,
0:34:00 you can use a workflow like this, right?
0:34:02 It lets you do a whole bunch of stuff in bulk.
0:34:04 And this will also work if,
0:34:06 let’s say I have like 20 tools listed here,
0:34:08 it’ll just one at a time go through the list
0:34:12 and follow that same process for every tool in the list.
0:34:14 – Yeah, I mean, that workflow is awesome.
0:34:15 I just, the whole time you’re talking there,
0:34:17 I’m sitting there thinking like, that’s really cool,
0:34:19 but it’s also very complicated.
0:34:21 And like, this is the kind of stuff
0:34:24 that agents are gonna like simplify, right?
0:34:25 Right, right.
0:34:26 And so, and I was wondering like,
0:34:28 okay, makes cool, is AppEars cool?
0:34:30 Do those sites still exist in the future?
0:34:31 Like when you have agents?
0:34:33 Like maybe that’s why Darmesh is making
0:34:34 that agent.ai thing.
0:34:38 Maybe he’s realized like those kind of sites
0:34:40 are gonna be simplified down into like,
0:34:41 okay, here are the agents.
0:34:42 And here’s like the kind of the instructions
0:34:44 for the agents.
0:34:46 And then you just hand the agents your API keys
0:34:49 and it does all that automatically.
0:34:51 – I mean, it’s a little bit more hands off
0:34:53 than I think I made it look
0:34:56 because I had to like turn off the automations
0:34:57 from running automatically
0:35:00 so that I could demo them on the video.
0:35:02 So literally what I’m doing when I add a new tool
0:35:07 to the website is I’m plugging in the URL of the website.
0:35:10 And then it just goes and does the rest, right?
0:35:12 It’s watching the Google sheet
0:35:15 for new information being added in like,
0:35:17 I don’t know, every half hour or every hour
0:35:18 or something like that.
0:35:19 It comes back, double checks the Google sheet
0:35:22 to see if there’s any new information on the Google sheet.
0:35:24 If there is, it goes through the list
0:35:25 and runs the automation.
0:35:29 So my role in all of this is literally just adding a URL
0:35:31 to the input URL on the Google sheet.
0:35:33 And then all of the rest of it happens
0:35:35 behind the scenes automatically.
0:35:38 So it’s sort of like I created my own little AI agent,
0:35:39 but you’re right.
0:35:42 I think in the future, I won’t have to build all that.
0:35:45 I’ll just have to tell like an AI or like a chat bot,
0:35:47 here’s what I want you to do.
0:35:49 And it will build all those automations
0:35:51 that I just sort of showed you the workflow for.
0:35:52 – Yeah, but then I’ll, like I said,
0:35:54 I’ll probably just ask you your API keys.
0:35:55 It’ll save that somewhere secure.
0:35:58 So it remembers it for the future
0:35:59 and it’ll just do all of that.
0:36:00 That’s gonna be so awesome.
0:36:01 – Yeah, yeah.
0:36:03 So share something less complex
0:36:06 ’cause that one was really in the weeds.
0:36:10 – Mine’s way less complex, so yeah.
0:36:14 So this is something I started doing recently.
0:36:16 So right now I’m kind of working,
0:36:17 I’ve had lore.com for a long time
0:36:18 and I’ve been trying to figure out what to do with it.
0:36:20 I’m still kind of trying to figure out
0:36:21 what to do with it beyond.
0:36:23 Like I like writing my newsletter,
0:36:24 but I feel like it’s such a great domain.
0:36:26 I should be doing something more with it.
0:36:29 And originally I bought the domain
0:36:31 ’cause I was trying to do the movie studio with Barry Osborn.
0:36:33 And so I’ve been interested in using it
0:36:35 for something more entertainment related for a long time.
0:36:38 So right now I’m doing this thing called Lore Labs.
0:36:40 I’m like, it’s kind of like a AI video agency
0:36:42 and a kind of community to help people
0:36:44 figure out how to create AI videos as well.
0:36:45 Still figuring it out.
0:36:46 So I’ve been working with these two great guys,
0:36:48 one of them who makes some of the best AI videos
0:36:49 I’ve ever seen.
0:36:50 I’m not announcing who he is yet,
0:36:52 but we’re collaborating on it.
0:36:56 And so we’ve got a slack full of different conversations
0:36:59 about like, okay, what should be our strategy, pricing?
0:37:01 What’s the initial strategy?
0:37:02 What are we gonna try if that doesn’t work?
0:37:04 Like, you know, lots and lots of conversations
0:37:06 to figure everything out.
0:37:08 And then we’re having weekly meetings.
0:37:10 And then one thing I realized that AI is really good at,
0:37:12 and I think other people could use it this way too,
0:37:14 is like using AI is almost like a meeting agenda.
0:37:16 Like, okay, you’re having a meeting,
0:37:18 you’ve been having all these conversations
0:37:20 through email or Slack or whatever.
0:37:23 Just throw them all in there into Claude
0:37:26 and you make a custom project for it or whatever.
0:37:29 And they say, hey, what are the actual like things
0:37:30 that need to be decided upon?
0:37:33 What are the like action items you need to do?
0:37:35 You know, key takeaways from the conversations.
0:37:37 And then just have it create the meeting agenda
0:37:39 automatically that way versus like,
0:37:40 well, you’re gonna sit there like 30 minutes
0:37:42 and like type up the meeting agenda
0:37:44 and have to re-remember what you actually talked about.
0:37:45 – Yeah, yeah, yeah.
0:37:46 – So that’s what I’m doing now
0:37:48 is I literally just copy and paste all of it in there.
0:37:50 And like I’ve told it, here’s how I want my meetings
0:37:53 structured, here’s the conversations,
0:37:56 figure out the agenda, and then send it out.
0:37:57 – That’s super smart.
0:37:59 So you’re basically, let’s say you’ve got some emails,
0:38:03 you’ve got some DMs, you’ve got some tweets,
0:38:06 I don’t know, you’ve got all of these various places
0:38:07 that you’ve had communication.
0:38:10 You just pull all of these pieces of content
0:38:13 in these communications you’ve had,
0:38:14 put them into Claude and say,
0:38:16 I’m about to meet with this person
0:38:20 based on everything I just put into your input.
0:38:22 How should we like, how should we keep
0:38:25 this meeting on track, right, an agenda for me?
0:38:26 – Yeah, totally.
0:38:28 And ideally there’s, you have like an agent
0:38:29 for that in the future, right?
0:38:30 – It actually goes out there and like,
0:38:32 yeah, we’re talking on email, we’re talking on Slack,
0:38:35 and you know, whatever, yeah, go and get that for us
0:38:36 and then actually create the invite.
0:38:37 That’s like the next step, right?
0:38:39 To actually do all of it.
0:38:41 But right now, yeah, I’m manually copying and pasting
0:38:44 and then just, it still saves me probably 20 minutes or so.
0:38:45 – Yeah.
0:38:46 – But yeah, ’cause some meetings it’s like,
0:38:48 okay, what did we talk about in the past?
0:38:50 Or what’s actually important to discuss?
0:38:52 You know, just tell the AI like what you want.
0:38:53 Like what’s, you know, how do you like meetings
0:38:55 to be structured?
0:38:55 – Yeah, well what’s–
0:38:58 – Feed the day to day and it’s pretty, it’s pretty good.
0:39:00 What’s even cooler too now is this is just
0:39:01 a brand new feature.
0:39:03 The week that we were recording that there was a new feature
0:39:05 in Google Meet where now we’ll summarize the meetings, right?
0:39:07 This has been in Zoom for a little while now,
0:39:10 but Google Meet just rolled out the feature where it will,
0:39:12 if you turn it on, it’s not always listening to your calls,
0:39:14 but if you turn it on, it will listen to your calls
0:39:16 and basically summarize the meeting for you
0:39:20 and make like a to-do list off of the meeting,
0:39:22 like an action item list kind of thing.
0:39:24 And so let’s say you need to have a second call
0:39:26 with that person, a follow-up call.
0:39:29 Well on that follow-up call, you just grab the meeting notes
0:39:32 from your last call, plug that in and say,
0:39:34 here’s what we talked about last time, you know?
0:39:36 – Yeah, yeah, yeah, what progress has been made,
0:39:40 what’s unresolved, you know, what do we need to discuss?
0:39:40 Yeah, it’s awesome.
0:39:42 – So really, really cool.
0:39:45 No, I think these are some awesome ideas
0:39:47 and I really like this format.
0:39:50 Like I love this format of just here’s some
0:39:51 of the cool ways we’re using them.
0:39:53 I think we’re gonna get to a point where we’re like,
0:39:54 all right, we’re running out of use cases
0:39:56 that we use ourselves,
0:39:58 but we’re constantly out there like fishing
0:40:00 for other use cases, we’re watching Twitter
0:40:03 and YouTube and Reddit and all these places
0:40:05 where people are sharing their cool AI use cases.
0:40:07 So I think it’ll be cool.
0:40:09 I think maybe next time we do one of these,
0:40:10 we’ll bring like a bunch of like,
0:40:13 here’s some cool workflows that we came across
0:40:14 that other people are using
0:40:15 that we think you should know about.
0:40:17 No, but this has been a fun discussion.
0:40:20 Once again, let us know your thoughts in the comments
0:40:21 wherever you’re listening to this.
0:40:23 If you’re on YouTube, let us know your thoughts.
0:40:25 If you have some cool use cases that you wanna share
0:40:27 and you want us to maybe shout ’em out on a future episode,
0:40:28 share them in the comments.
0:40:32 That might be a really great way to discover new workflows.
0:40:33 If we talk about your workflow,
0:40:35 we’ll shout out your name on the episode
0:40:38 and let everybody know that that came from you.
0:40:40 But I think we’re probably gonna do more of these episodes.
0:40:42 So let us know what you think of them in the comments
0:40:44 either on YouTube or Spotify
0:40:46 or wherever you’re listening to podcasts.
0:40:48 And if you wanna make sure more of these podcasts
0:40:50 show up in your feed, make sure you like and subscribe.
0:40:51 We really, really appreciate it.
0:40:53 It makes us feel a little warm and fuzzy.
0:40:55 And thank you so much for tuning in.
0:40:57 – Yeah, thank you so much.
0:40:59 (upbeat music)
0:41:02 (upbeat music)
0:41:05 (upbeat music)
0:41:07 (upbeat music)
0:41:10 (upbeat music)
0:41:12 you
0:41:14 you
0:00:04 FutureTools.io.
0:00:06 [LAUGHTER]
0:00:09 We’re figuring out cool use cases in real time
0:00:10 on this episode right now.
0:00:12 Yeah, yeah.
0:00:15 Five store– hey, it licked the FutureTools, yeah.
0:00:18 I’m telling you, there’s like an SEO hack right now.
0:00:21 [MUSIC PLAYING]
0:00:23 Hey, welcome to the Next Wave podcast.
0:00:24 I’m Matt Wolf.
0:00:26 And once again, I’m here with Nathan Lanz.
0:00:30 And today, we’re talking once again about AI use cases.
0:00:34 And I feel like a lot of people sort of know the surface level
0:00:36 uses of AI, right?
0:00:39 You kind of know how to go have conversations with JetGPT or Claude
0:00:42 or maybe make an image with mid-journey.
0:00:44 But so many people are just sort of looking
0:00:45 at the tip of the iceberg.
0:00:47 And that iceberg goes so deep.
0:00:49 There are so many cool things that you
0:00:51 can do with these AI tools.
0:00:53 And in this episode, we’re going to deep dive
0:00:56 into some of the more interesting use cases
0:00:58 that you can use AI for.
0:01:00 Some of them fairly basic and easy to implement.
0:01:04 And one of them is a little more complex and sort of a deep dive
0:01:07 into literally how I’m using AI to run my business.
0:01:10 I sort of pull back the curtain and show you
0:01:13 the whole process of what I’m doing to run my FutureTools
0:01:14 website using AI.
0:01:16 And we’re going to get to all of that in this episode.
0:01:18 I think you’re going to get a lot of value out of it
0:01:21 and learn a whole bunch of new ways to use AI.
0:01:25 So let’s just go ahead and jump right into it.
0:01:27 [MUSIC PLAYING]
0:01:30 When all your marketing team does is put out fires,
0:01:31 they burn out fast.
0:01:34 Sifting through leads, creating content for infinite channels,
0:01:38 endlessly searching for disparate performance KPIs,
0:01:39 it all takes a toll.
0:01:43 But with HubSpot, you can stop team burnout in its tracks.
0:01:45 Plus, your team can achieve their best results
0:01:46 without breaking a sweat.
0:01:49 With HubSpot’s collection of AI tools, Breeze,
0:01:52 you can pinpoint the best leads possible.
0:01:55 Capture prospects attention with click-worthy content
0:01:58 and access all your company’s data in one place.
0:02:00 No sifting through tabs necessary.
0:02:02 It’s all waiting for your team in HubSpot.
0:02:06 Keep your marketers cool and make your campaign results hotter
0:02:06 than ever.
0:02:09 Visit hubspot.com/marketers to learn more.
0:02:12 [MUSIC PLAYING]
0:02:14 Just go ahead and kick it off.
0:02:19 I’ll share a fairly simple, basic one that I use.
0:02:22 This one is using Claude.
0:02:23 I use Claude a lot.
0:02:25 So a lot of my workflows use Claude.
0:02:28 I love making projects inside of Claude.
0:02:31 But one of these little projects that I made
0:02:33 is called the News Summarizer.
0:02:37 So if you watch my other YouTube channel, every Friday,
0:02:39 I put out an AI news video where I break down,
0:02:43 like, here’s all of the latest AI news for the week.
0:02:46 And this is pretty much how I come up
0:02:50 with what I’m going to say about the news, for the most part,
0:02:52 is I’ve created this new summarizer here.
0:02:55 And I’ll go ahead and show you my system prompt.
0:02:58 I’ll read it off as well for anybody that’s listening.
0:03:01 But I have this system prompt here that says,
0:03:04 I will upload or paste in an article, PDF, or video
0:03:05 transcript.
0:03:07 Your job is to respond with bullet points
0:03:10 with the following information.
0:03:12 Summarize the content into bullet points,
0:03:16 making it easy to understand the concepts or ideas presented.
0:03:19 Tell me what we can do now as a result of this information
0:03:20 that we could not do before.
0:03:22 What makes this novel?
0:03:24 Tell me how I should report on this in the news.
0:03:27 If I was to give a news broadcast and explain why
0:03:30 this information was important, what should I tell the viewers
0:03:33 so that the most amount of people can understand it,
0:03:35 the format of the response should be,
0:03:37 and then I give a little format here, summary, bullet,
0:03:40 bullet, bullet, what makes it novel, bullet, bullet, bullet.
0:03:43 And then how should I describe this news or concept
0:03:45 in a simple way that anyone can understand
0:03:46 when I report on it?
0:03:49 If it makes sense, use an analogy.
0:03:51 So that’s the whole system prompt there.
0:03:53 And basically, what I’m telling it to do
0:03:54 is I’m going to give you content.
0:03:56 It might be a PDF, might be a document,
0:03:58 might be a copy and pasted news article.
0:04:02 Break it down for me so I can explain it really, really easily.
0:04:05 And one of the things that’s been sort of circulating
0:04:07 in the world of AI news this week–
0:04:09 I’m sure you’ve probably seen this, Nathan–
0:04:10 is that doom thing, right?
0:04:15 The AI doom where it’s sort of generating every single frame.
0:04:17 So as somebody clicks the forward button,
0:04:20 it generates the next frame in doom.
0:04:21 If they click the shoot button,
0:04:23 it automatically generates the next frame.
0:04:25 There’s no underlining–
0:04:26 Game engine.
0:04:28 Yeah, no game engine underneath it.
0:04:31 It’s all being generated as they’re playing.
0:04:34 So I have that PDF, and I can drag that PDF.
0:04:37 This is literally the research report
0:04:40 right off of archive.org that explains
0:04:43 the technical concept of how this works.
0:04:46 And if you’re looking at my screen right now,
0:04:47 if you’re not looking at my screen,
0:04:49 you can look at it over on YouTube.
0:04:52 But I just dropped the PDF right here into Claude,
0:04:54 and I don’t even add a prompt.
0:04:56 All of my prompting was done in that system prompt
0:04:57 that I just read.
0:04:59 And if I just hit Enter on this,
0:05:01 it’s going to read that PDF for me
0:05:04 and then break it down with all of the information
0:05:07 that I asked it to give me in that system prompt.
0:05:11 So right now, it’s generating all of that text for me here.
0:05:12 And here’s what it says.
0:05:15 Here’s a summary and analysis of the research paper.
0:05:17 Summary– the paper introduces game engine,
0:05:19 a neural model that can simulate the classic game
0:05:23 doom in real time at over 200, at over 20 frames per second,
0:05:25 on a single TPU.
0:05:28 And it summarizes what the article’s about.
0:05:31 And then the next section, what makes this novel interesting.
0:05:33 This is the first neural model capable
0:05:36 of simulating a complex 3D game like doom in real time
0:05:38 with high visual quality.
0:05:40 It demonstrates the neural models
0:05:42 can potentially replace traditional game engines
0:05:45 for interactive 3D environments, et cetera, et cetera.
0:05:47 And then it goes on to say, here’s
0:05:49 how to describe this news simply, right?
0:05:54 So now you can start to see how on my AI news videos–
0:05:56 yes, I put a lot of work into them.
0:05:58 But AI supports the hell out of me
0:06:00 for making those news videos.
0:06:03 So here’s what it says for how to describe this news simply.
0:06:05 Imagine if instead of programming
0:06:07 every detail of a video game, we can just
0:06:11 show an AI system examples of the game being played.
0:06:14 And it can learn to recreate the entire game on its own.
0:06:16 That’s essentially what researchers
0:06:19 have achieved with a classic game called doom.
0:06:21 They trained an AI to watch gameplay videos
0:06:24 and then reproduce the game so accurately
0:06:26 that even humans have trouble telling
0:06:29 the difference between the AI version and the real thing.
0:06:33 This could revolutionize how video games are made in the future,
0:06:34 potentially making it much faster and easier
0:06:37 to create complex 3D worlds and gameplay.
0:06:40 It’s like teaching an AI to be a game designer and programmer
0:06:42 all in one, right?
0:06:45 So that’s like one of my most used workflows.
0:06:47 And this is very relevant.
0:06:49 We’re recording this on a Thursday.
0:06:52 I record my news videos for YouTube on a Thursday.
0:06:57 So I was in here actually using this Claude project earlier
0:07:01 today to start to break down my news videos.
0:07:04 Now, every piece of news that I put in those news videos
0:07:05 doesn’t need this process, right?
0:07:08 Like some of the news is just sort of self-explanatory.
0:07:10 But every once in a while, I’ll come across a research paper
0:07:14 and I’m like, I want to go a little bit more in-depth on this.
0:07:18 I really want to make sure that the viewers understand
0:07:20 what this means and this is how I do that.
0:07:22 I’ll just plug in the research paper
0:07:23 and it’s going to simplify it for me
0:07:28 and then tell me how I should break it down as a piece of news.
0:07:31 I’m curious, how much time do you think this saves you?
0:07:34 There’s a time savings, but there’s also a mental savings, right?
0:07:37 So it probably saves me a good 15 or 20 minutes
0:07:41 of actually going through and reading all of the concepts
0:07:43 within the research paper.
0:07:45 But at the same time, these research papers
0:07:46 are often over my head.
0:07:49 Like let’s be honest, once they start breaking down math
0:07:52 and there’s letters within the math equation
0:07:58 and it’s like I X over 15% plus 32 AD plus 6.
0:08:00 I’m like, you lost me with I, right?
0:08:02 Like the very first letter in there, right?
0:08:07 So like the mental load that it takes off of my plate
0:08:10 of like me reading this research paper like three times
0:08:13 to try to understand what the hell it’s actually saying
0:08:15 is probably the bigger savings
0:08:17 than the actual time savings for me.
0:08:18 – Yeah, I feel like I should be using this
0:08:20 for my newsletter probably.
0:08:23 I do wonder like it feels like, you know,
0:08:25 maybe a lot of our listeners, they’re probably not YouTubers
0:08:27 or they’re probably like, okay, how do I use this?
0:08:29 You know, it feels like you could use this
0:08:30 in a lot of different parts of business, right?
0:08:33 Like even if it was like contracts or whatever,
0:08:36 any kind of complicated paperwork, papers,
0:08:37 almost anything, any kind of PDF,
0:08:40 you could make a project in Claude or even using,
0:08:42 you know, a Chesapeake T, maybe Claude’s kind of better
0:08:43 for this right now though.
0:08:45 You could create a custom project and kind of tell it like,
0:08:47 okay, here’s the kind of data to expect
0:08:49 and here’s kind of like what I want out of the data.
0:08:51 Here’s the kind of stuff that’s actually important to me
0:08:53 and then just feed the data in
0:08:56 and like it does an amazing job of just like spitting out
0:08:57 stuff that’s actually important for you.
0:09:00 – Yeah, yeah, I mean, I originally came up
0:09:02 with this concept for my newsletter.
0:09:04 Now I have help, I have extra writers and editors
0:09:07 helping write the newsletter for me,
0:09:11 but I originally generated this as a custom GPT
0:09:14 over in chat GPT, but I actually like the way
0:09:15 Claude does it better.
0:09:17 I like Claude’s projects better.
0:09:18 I like the way it’s organized better.
0:09:21 I think Sonnet 3.5 gives me better outputs
0:09:24 than what the current model of GPT-4 gives me.
0:09:28 So I sort of re-implemented it over in Claude
0:09:29 to get the same thing,
0:09:32 but now I’ve started using it for videos
0:09:33 instead of my newsletter.
0:09:35 But honestly, like this can just be used
0:09:39 for anything you want explained more simply, right?
0:09:42 Like if you come across some sort of article that you’re like,
0:09:44 that seems like it’s pretty interesting,
0:09:44 but it’s over my head.
0:09:47 Like how do I understand this?
0:09:49 Like how do I just like grasp it?
0:09:52 Stuff like this really helps to just like grasp concepts.
0:09:54 Like use some analogies, give me bullet points,
0:09:56 explain it as simply as possible.
0:09:58 – Yeah, it feels like opening, I really dropped the ball
0:10:02 with like not having something like the Claude projects.
0:10:03 Like the projects are so useful.
0:10:05 I think like you first learned about the projects
0:10:07 or how to use them like on this podcast, right?
0:10:09 Like I told you like, oh, that just came out
0:10:10 and you’re like, oh cool, I heard about that.
0:10:11 Like what is it?
0:10:14 And like, yeah, here’s how you use it.
0:10:16 – I think so, the first time I ever used them,
0:10:18 I think you sort of like I knew artifacts
0:10:20 and I was using the artifacts thing.
0:10:21 And I was like, I saw projects,
0:10:23 but in my mind projects were just like a folder
0:10:24 to store stuff in.
0:10:27 I didn’t realize it had the custom instructions
0:10:30 and the ability to upload additional like information
0:10:34 that it would sort of retrieve when you’re prompting.
0:10:36 And I didn’t know it had that extra stuff yet
0:10:38 when we talked about it on the podcast.
0:10:40 And now I’m like, this is the best thing ever.
0:10:42 I’m in here like all day.
0:10:45 – Yeah, I mean, I think most listeners
0:10:47 probably don’t even understand even like custom instructions
0:10:48 or that chat to BT has that.
0:10:51 They were the kind of the first to have custom instructions
0:10:54 where, you know, and I feel like sometimes
0:10:56 when people are talking about the quality of these models,
0:10:59 I do wonder like, have they ever used custom instructions?
0:11:01 Like maybe they haven’t.
0:11:04 Like I probably should show my custom instructions
0:11:06 at some point, but I feel like it gives me
0:11:07 a lot better results from the model.
0:11:11 Like I tell it to like, okay, be less politically correct,
0:11:14 be more straightforward, tell me the truth.
0:11:15 You know, and all these kinds of things.
0:11:17 And I kind of tell like what level of detail do I want?
0:11:20 And you can give it, you know, give the models
0:11:22 that those kinds of custom instructions.
0:11:24 Now the thing that Claude’s doing better than OpenAI
0:11:25 is that you can have different projects
0:11:27 with different custom instructions
0:11:29 based on what you’re trying to accomplish.
0:11:31 – Well, you can technically do it in chat GPT.
0:11:33 You just create a GPT for each thing.
0:11:36 ‘Cause you can give every GPT its own custom instructions.
0:11:38 The problem I was having with chat GPT
0:11:42 was I actually built a GPT to do this kind of thing for me.
0:11:46 And it worked really, really well for several weeks.
0:11:48 And then one day I went in there and tried to use it
0:11:51 and it wasn’t formatting the information correctly.
0:11:53 And it wasn’t giving me analogies
0:11:55 like I was asking it to just like one day
0:11:57 it started ignoring my custom instructions.
0:11:59 And I’m like, what happened?
0:12:00 And I went and opened the custom instructions
0:12:02 and that GPT, nothing changed.
0:12:03 They were the same.
0:12:07 It just felt like chat GPT got dumber one day.
0:12:09 And so I went and tested over on Claude and I’m like,
0:12:14 okay, Claude is out custom GPT-ing GPT, you know?
0:12:15 – Yeah, yeah.
0:12:17 I mean, OpenAI has done this a lot
0:12:19 where they’ve been making changes so quickly.
0:12:20 And sometimes they’re great.
0:12:22 And sometimes like, oh yeah, they broke something.
0:12:24 They came out with that memory feature
0:12:25 which they barely talked about.
0:12:26 And I thought it was an amazing feature
0:12:29 where it’s like, it’ll just start adding stuff to memory.
0:12:30 I’m like, oh, this is amazing.
0:12:32 The model is actually remembering stuff.
0:12:33 And then, oh, I can actually add,
0:12:35 I can manually add stuff to the memory.
0:12:36 And then now it’s gonna remember it forever.
0:12:37 That’s awesome.
0:12:39 And then you find out like there’s like a major limit
0:12:41 to the memory, and not only that,
0:12:43 but it’ll actually go in and like remove things
0:12:44 from the memory.
0:12:45 – Yeah, yeah, yeah.
0:12:47 – It’ll be like, oh, and even if you don’t
0:12:49 touch remember something, you might type something to it.
0:12:51 And it’s like, oh, it wants to remember that now.
0:12:52 And it says memory updated.
0:12:54 And then like literally you can go check memory
0:12:55 and then sometimes that means
0:12:56 it actually removed something else.
0:12:58 – Yeah, it’ll remember stuff that you’re like,
0:13:00 why are you remembering that specifically?
0:13:02 That doesn’t need to be remembered.
0:13:03 – Right, right, yeah.
0:13:04 I’ll be asking it for advice
0:13:05 on something I’m doing in business
0:13:06 or whatever and life, whatever.
0:13:09 And like it’ll be like memory updated, like what?
0:13:10 What do I do?
0:13:11 (laughing)
0:13:13 (upbeat music)
0:13:14 – We’ll be right back.
0:13:17 But first I wanna tell you about another great podcast
0:13:18 you’re gonna wanna listen to.
0:13:20 It’s called Science of Scaling,
0:13:21 hosted by Mark Roberge.
0:13:24 And it’s brought to you by the HubSpot Podcast Network,
0:13:28 the audio destination for business professionals.
0:13:30 Each week host Mark Roberge,
0:13:32 founding chief revenue officer at HubSpot,
0:13:34 senior lecturer at Harvard Business School
0:13:36 and co-founder of Stage 2 Capital,
0:13:39 sits down with the most successful sales leaders in tech
0:13:42 to learn the secrets, strategies, and tactics
0:13:44 to scaling your company’s growth.
0:13:46 He recently did a great episode called
0:13:50 How Do You Solve for a Siloed Marketing in Sales?
0:13:52 And I personally learned a lot from it.
0:13:54 You’re gonna wanna check out the podcast,
0:13:57 listen to Science of Scaling wherever you get your podcasts.
0:14:00 (upbeat music)
0:14:02 – What’s your first one you wanna share?
0:14:04 – Yeah, mine is kind of a personal one.
0:14:07 It’s also, I’m gonna use cloud projects for it.
0:14:10 My entire marriage and relationship in Japan,
0:14:12 like when I first moved to Japan,
0:14:14 my Japanese is very basic.
0:14:16 I can’t have full blown conversations with anyone.
0:14:19 I can have like, I can say small little words
0:14:21 enough to kind of get some basic points across.
0:14:25 But then I started using chat to BT for translation.
0:14:27 And I did it through like basically the custom instructions
0:14:29 where I would like say like,
0:14:31 ’cause obviously you can tell chat to BT
0:14:32 to translate something for you,
0:14:35 but it’s kind of tedious when you’re doing it all the time.
0:14:36 Like, yeah, you know, if you’re like,
0:14:38 yeah, if you’re translating a page
0:14:39 or you copy and paste everything
0:14:41 and you explain what you want,
0:14:43 but that gets annoying
0:14:45 if you’re actually in a conversation with someone, right?
0:14:47 So basically I would use custom instructions
0:14:48 to make like short hands, like saying,
0:14:53 “Hey, if you see, if I put T at the beginning of a sentence,
0:14:55 that means I want this translated.
0:14:58 I don’t wanna tell you, hey, please translate all of this.”
0:14:59 And so that’s what they haven’t seen people do much
0:15:02 is you can actually make a lot of cool short hands yourself
0:15:04 with chat to BT or with Claude.
0:15:06 And I don’t really see anyone talking about that,
0:15:08 but it’s really, it’s nice.
0:15:08 You can make short hands.
0:15:10 Like here’s this letter or whatever,
0:15:11 and this is what it means.
0:15:13 And this is what I want to accomplish now.
0:15:16 And then you just put that the beginning of a sentence
0:15:18 and now it’ll do what you want.
0:15:20 So I use that for translations.
0:15:22 I want something translated, put a T, then I put it.
0:15:25 It knows to translate it.
0:15:27 And then similar to like how you give, you know,
0:15:28 with your YouTube channel,
0:15:30 you tell it all this other stuff you want it to do.
0:15:32 I do the same thing with the translations.
0:15:34 Like, hey, when I’m translating something to her,
0:15:37 like maybe give me cultural feedback too,
0:15:38 ’cause Japan’s quite a bit different.
0:15:40 So don’t just change what I’m saying,
0:15:41 but do give me feedback.
0:15:42 Like if like, maybe for Japanese people,
0:15:45 this is very odd to say something like this, right?
0:15:47 Like give me cultural feedback,
0:15:49 but don’t change it without asking me, right?
0:15:51 Like I tell all this kind of stuff.
0:15:52 So don’t change it without asking me.
0:15:54 Also show the English underneath.
0:15:57 So I know that you didn’t change anything.
0:15:59 ‘Cause occasionally when you tell it to translate,
0:16:01 it does change things without telling you.
0:16:05 So you should always get a version in your own language
0:16:06 to kind of double check.
0:16:07 But also I’ve used it for more,
0:16:09 like people were saying when we did the episode before,
0:16:10 when I talked about this, they were like,
0:16:12 “Oh, so you’re literally not gonna learn Japanese.”
0:16:14 I’m like, “No, I’m trying to learn Japanese too.”
0:16:16 Like I’m actually trying to do both at the same time.
0:16:17 So when I have it translate,
0:16:19 I also have it break things down for me,
0:16:21 which has been one of the coolest things.
0:16:25 Like Japanese has three different types of characters,
0:16:26 character systems for their language.
0:16:28 It’s like, makes it very complicated.
0:16:31 Yeah, you’re gonna Katakana and Kanji.
0:16:33 Hiragana being the most basic one.
0:16:36 So I know Hiragana very well.
0:16:38 I know Katakana kind of okay and Kanji.
0:16:40 Oh my God, that takes like a long, long time to learn.
0:16:42 So what I have it do is basically,
0:16:44 you have my translate something.
0:16:46 I’m like, okay, what are the key words in that sentence
0:16:47 that I should learn?
0:16:51 And then like break them down for me in Hiragana.
0:16:53 So it’ll be like, okay, you made this sentence
0:16:55 and then here’s like three bullet points.
0:16:57 I’ll tell you, I don’t wanna learn like five words at a time.
0:16:58 Make it like three.
0:16:59 I’ll learn like three words.
0:17:01 What are the three most important words in that sentence?
0:17:03 And then break them down for me
0:17:04 and then show me the Hiragana
0:17:06 ’cause I can actually read that in Japanese.
0:17:08 And it’s made it a lot easier for me to learn the language.
0:17:10 Just like, okay, here’s that word.
0:17:11 I just was wanting to translate
0:17:14 and now break it down for me and teach it to me.
0:17:16 The next level, I mean, ideally in the future,
0:17:18 like you can like the memory be really good, right?
0:17:20 And you could like have the model help you learn.
0:17:22 Like, oh, Nathan needed to learn these words
0:17:24 or he learned these before, test him on it
0:17:25 to see if he actually learned it.
0:17:27 But unfortunately, they’re not really like good enough
0:17:28 to like remember all that.
0:17:31 – Yeah, but I feel like that’s probably coming soon.
0:17:31 – Yeah, yeah.
0:17:35 No, I noticed Riley Brown, you know Riley Brown, right?
0:17:38 He does like short form content around AI.
0:17:40 I saw him use a similar trick
0:17:41 because he’s doing a lot of stuff
0:17:44 where he’s using Claude to help him code up his,
0:17:47 his like, he’s trying to make like an app a day
0:17:48 for 30 days or something like that, right?
0:17:49 – Yeah, yeah, yeah.
0:17:51 – And one thing Claude tends to do
0:17:53 when you’re asking it to write code
0:17:57 is it will write just the code that needs to be swapped out
0:17:59 as opposed to rewriting the entire thing.
0:18:02 So let’s say just for like a real basic example,
0:18:05 it writes out a whole like HTML webpage for you.
0:18:09 And then you say, oh, the font at the top needs to be changed.
0:18:11 Well, what it’ll do is it’ll say, okay,
0:18:13 here’s the code that needs to be changed.
0:18:14 It’ll just give you like three lines of code.
0:18:17 And you’re like, okay, if I don’t know anything about code,
0:18:19 I don’t know where that code is supposed to go.
0:18:20 Like, what am I supposed to replace?
0:18:22 – Well, maybe if you know the code,
0:18:23 I mean, I know how to code decently.
0:18:24 Like it’s still annoying.
0:18:26 So I feel kind of dumb now for not having done this.
0:18:26 So I get where you’re going.
0:18:28 Like he has short hands to tell it.
0:18:29 And he makes sure to tell it,
0:18:31 like give me the entire code, probably.
0:18:33 – He has a short hand that’s like,
0:18:35 I don’t remember what his actual short code is,
0:18:38 but it might be like CCC or something like that.
0:18:41 He’ll put like CCC at the end of one of his prompts,
0:18:44 but it’s a short code to like remind it,
0:18:46 give me the entire code.
0:18:49 Don’t just give me like what needs to be replaced, right?
0:18:50 And I thought that was really smart.
0:18:51 I started implementing that
0:18:53 when I start playing around with code as well.
0:18:55 I always tell it, give me the entire code.
0:18:58 Don’t give me just what I need to replace.
0:18:59 And like the little short code thing
0:19:01 saves quite a bit of time on that aspect.
0:19:02 – I need to be doing that
0:19:05 ’cause I’ve definitely used Claude for coding.
0:19:07 And like, and I’ve done that many times
0:19:08 where like it’ll give me the small little snippet.
0:19:09 I’m like, damn it.
0:19:10 Now I gotta like read through the code
0:19:12 and it’ll take like 30 seconds or a minute.
0:19:13 Like give me the whole thing.
0:19:14 – Yeah, yeah, yeah.
0:19:16 – And I always wonder why it doesn’t do that.
0:19:17 Maybe because it’s like, you know,
0:19:19 it’s saving energy or I have no idea why they’re doing it.
0:19:21 Like saving compute or something.
0:19:22 I don’t know, but it is annoying
0:19:23 where it’ll just like give you a small snippet.
0:19:25 It’s like, yeah.
0:19:27 – Yeah, and you can even, the funny thing is,
0:19:30 I even have it in my system prompt of like,
0:19:33 when I ask you to generate code, generate the entire code.
0:19:35 Don’t just tell me what needs to be replaced.
0:19:37 And it literally always ignores that.
0:19:40 Like you need to, for whatever reason, put it in the prompt.
0:19:41 It doesn’t work in the system prompt.
0:19:43 It’s so annoying.
0:19:46 I’m sure they’ll like, that’ll get patched up soon,
0:19:47 but for whatever reason,
0:19:49 they ignores that part of the system prompt.
0:19:51 So the next thing I’ll share is,
0:19:54 I’m gonna share a little perplexity tip.
0:19:55 This one will be pretty quick here.
0:19:57 One thing that I think is pretty overlooked
0:20:01 inside of perplexity, but I feel is pretty powerful,
0:20:04 is they have a feature here,
0:20:07 where if I go to my library
0:20:09 and go to this page feature here,
0:20:12 this page feature inside of perplexity,
0:20:16 will essentially create like a little mini like Wikipedia
0:20:18 on any topic you want it to create, right?
0:20:21 So if I hear about like a concept
0:20:22 that I don’t understand very well,
0:20:24 but I wanna learn more about it,
0:20:28 I’ll come to this page feature inside of perplexity
0:20:29 and have it actually generate
0:20:31 like a little mini Wikipedia for me.
0:20:35 So for example, let’s say I wanna learn quantum computing,
0:20:36 right?
0:20:37 Like I don’t understand quantum computing.
0:20:39 I could literally just type quantum computing,
0:20:42 hit the little button to tell it to go.
0:20:44 And then it will literally just start generating
0:20:46 essentially a Wikipedia page.
0:20:49 It explains everything I can possibly want to know.
0:20:50 Well, probably not everything.
0:20:52 There’s probably quite a bit of depth of quantum computing.
0:20:54 But I can explain pretty much everything,
0:20:56 a surface level person might wanna know
0:20:58 about quantum computing.
0:21:00 So you can see quantum speedups and database searches,
0:21:03 quantum algorithms for optimization problems,
0:21:07 quantum error mitigation techniques.
0:21:09 And it’s just got like this whole page now
0:21:11 with sources that I can click into
0:21:14 and click over to the websites to learn more.
0:21:16 And maybe there’s something that it’s missing,
0:21:21 like let’s say overlap of quantum computing and AI.
0:21:27 I’ll hit plus and now it’s just gonna add a little section
0:21:31 that tells me all about the overlap of quantum computing
0:21:34 and AI and how they’re related and things like that.
0:21:37 So this to me has been really, really powerful
0:21:39 whenever I sort of get a new concept in my mind
0:21:43 that I’m like, I need to dig into that a little bit more.
0:21:46 Like this tool is really powerful for just like diving deeper
0:21:49 and deeper and deeper because I can go here
0:21:50 and just tell it to insert another section
0:21:52 and add more about it.
0:21:55 Quantum speedups and database searches.
0:21:57 Okay, there’s not enough info there for me.
0:21:58 I’ll click more.
0:22:00 That doesn’t do what I thought it did.
0:22:02 I thought it was gonna actually add more to it,
0:22:03 but it doesn’t.
0:22:04 But I can just come down here.
0:22:07 – It saves this as an actual page too, right?
0:22:08 – Yeah, it saves it as an actual page.
0:22:10 So if I come back to my library here,
0:22:13 I have a tab here for threads,
0:22:16 which are like the questions I’ve asked for Plexity.
0:22:19 And then I have pages here and you can see like,
0:22:20 it’s right there.
0:22:23 I can get right back to this and it’s also shareable.
0:22:25 I can click publish up here in the top right
0:22:27 of this perplexity page.
0:22:29 And when I publish it, it’s gonna give me a link
0:22:32 that I can share and now I can send this link
0:22:33 to anybody I want.
0:22:36 And they can see my quantum computing revolution page
0:22:41 that I had generated for me here in a matter of a minute.
0:22:43 So I found that to be really powerful.
0:22:46 – Yeah, I saw some people on SEO Twitter talking
0:22:48 about this saying that like actually these pages
0:22:50 are ranking quite well right now.
0:22:51 So for listeners, that might be something,
0:22:53 I haven’t tested it myself.
0:22:54 I don’t know if it works well,
0:22:57 but it definitely probably if you’re relying on SEO,
0:22:59 maybe it’s a good strategy to test.
0:23:01 – How would we get our website in here though?
0:23:04 Let’s say, so if I get to this version where I can edit,
0:23:07 I want to try to see if I can like inject my link.
0:23:12 Quantum computing and futuretools.io.
0:23:16 (both laughing)
0:23:18 Let’s see if it’ll actually generate content
0:23:22 around that relationship.
0:23:24 I’ve sort of, hey, it licked the futuretools.
0:23:25 Yeah.
0:23:28 – I’m telling you, I’ve been hearing people are saying
0:23:30 like there’s like a kind of a SEO hack right now.
0:23:32 I’m not sure I haven’t tested myself,
0:23:35 but people are saying this is like a great way
0:23:38 to like get links to your site and maybe some traffic.
0:23:40 I’m not sure if the back links or follow links or not.
0:23:42 – We’re figuring out cool use cases in real time
0:23:43 on this episode right now.
0:23:46 – Yeah, yeah, they really nailed it.
0:23:47 I’ve actually, I mean, I read about this,
0:23:48 but I haven’t actually tried it.
0:23:49 I should have tried it.
0:23:50 It looks so cool.
0:23:52 It’s like cooler than I expected.
0:23:55 My buddy, Jude, who was like an early guy at YC,
0:23:56 he did something similar,
0:24:00 how he tried to called Golden back a few years ago.
0:24:03 And I think they were mostly relying on manual
0:24:04 and they tried to shift to AI.
0:24:06 And it was like right when AI was just starting
0:24:08 and just it ended up not going,
0:24:10 I think he sold the company for a decent amount,
0:24:11 but it didn’t get as big as he hoped.
0:24:13 And he was trying to do the same kind of thing
0:24:14 ’cause actually I was sitting there thinking like,
0:24:16 hey, maybe lore.com is a great name for that too.
0:24:18 Or something like this, you know, the same kind of thing,
0:24:20 like the lore behind some kind of,
0:24:22 maybe like more like fantasy and consumer kind of stuff
0:24:23 versus like businesses, but.
0:24:24 – No, it’s really cool.
0:24:28 It’s a great tool for just sort of deep diving on a concept.
0:24:29 And you could just keep adding new sections
0:24:34 and deep diving into various topics on this,
0:24:35 you know, on this page.
0:24:38 I can even like select and highlight certain areas
0:24:41 and it will sort of extrapolate on certain areas
0:24:42 that I highlight as well.
0:24:44 So just a cool deep dive tool.
0:24:46 – You could, yeah, you could use it as a learning tool,
0:24:47 but you also, you probably could use it as a way
0:24:50 to communicate things to your team or something,
0:24:52 something they should know about or to learn.
0:24:53 You could kind of curate,
0:24:54 like here’s all the information on a topic
0:24:57 that’s actually relevant that you should learn.
0:25:00 – Yeah, I was building out a page not too long ago
0:25:03 that was all about YouTube growth strategies, right?
0:25:05 And I just kept on having it add new sections
0:25:06 about ways to grow on YouTube.
0:25:09 And it just kept on finding more information
0:25:11 about YouTube growth strategies
0:25:14 and adding new sections related to how to grow on YouTube.
0:25:15 And I’m like, this is so cool.
0:25:17 Like I could just get lost in this.
0:25:18 – That’s so cool.
0:25:19 Yeah, I’ve read about it.
0:25:20 I thought it was cool.
0:25:23 And I’m like, no, like, yeah, I gotta go play with this.
0:25:25 I’m probably gonna do that after the show.
0:25:26 – Yeah, yeah.
0:25:30 – Cool, so this one is from our friend Billawall Sidhu,
0:25:32 who does the TED AI podcast.
0:25:33 – Yep.
0:25:34 – I thought this was kind of cool.
0:25:37 He’s showing how you can use Idiogram 2.0,
0:25:38 like the new, one of the new AI art tools
0:25:42 that we recently talked about to make YouTube thumbnails.
0:25:43 It’s not perfect.
0:25:44 I mean, I would say it’s probably not as good
0:25:45 as like the ones we’re currently making
0:25:48 or on your YouTube channel, but it’s really good at text.
0:25:51 Like used to, that was a thing that AI art really struggled
0:25:53 with, it was like getting text on an image.
0:25:55 And now apparently like they’ve just like
0:25:58 completely nailed text, like it works.
0:26:00 – Idiogram’s the best at text for sure.
0:26:03 Nothing close to what Idiogram can do.
0:26:06 – Yeah, and he said apparently, you know,
0:26:08 actually here tags you here.
0:26:09 Apparently, you know, they’re also,
0:26:10 they’re working on likeness.
0:26:12 So you probably can do, you know,
0:26:14 kind of like how you do where you have like AI art
0:26:17 that’s similar to your face and use that as a thumbnail.
0:26:19 Like apparently that’s in the works.
0:26:21 So probably the next few months,
0:26:22 you’ll be able to make like, you know,
0:26:25 your own thumbnails just using Idiogram
0:26:26 versus like having to pay someone
0:26:28 or spend a bunch of time in Photoshop or whatever.
0:26:30 So I think for people who are doing YouTube
0:26:32 or have YouTube channels or, you know,
0:26:33 this is a great use case.
0:26:35 And also maybe even for like presentations, right?
0:26:37 Like if you’re making a corporate presentation,
0:26:40 like people love to have like fancy looking art
0:26:42 with text over it in presentations.
0:26:45 It makes your text, your presentations to the next level.
0:26:45 I think people right now
0:26:47 should be using Idiogram for that.
0:26:49 – Yeah, does he break down a workflow there?
0:26:52 Is it like, is there like a specific process
0:26:54 he’s using to get those results?
0:26:55 Or is he basically just saying like,
0:26:58 “Hey, Idiogram is great at this.”
0:26:59 – As far as I can tell,
0:27:00 he just think Idiogram is great at this.
0:27:02 I don’t think there’s any like, yeah.
0:27:04 – Yeah, ’cause I can see the thumbnails that it made
0:27:05 and it looks good.
0:27:07 It almost looked like a type of thumbnail
0:27:11 you might see on like a like an NPR
0:27:12 sort of YouTube channel or something like that.
0:27:13 – Yeah.
0:27:16 – But I’m just curious what the prompts would be
0:27:17 to get it looking like that, right?
0:27:19 Because the way it’s sort of,
0:27:21 the text is justified to the left
0:27:24 in the bottom left corner of the images.
0:27:27 Like on the reality bending 3D captures,
0:27:30 the word captures is highlighted in blue.
0:27:32 Like I wonder if there was like some prompt tricks
0:27:33 that he used to get that.
0:27:37 Or if he just like said, make an image that says this.
0:27:40 Yeah, I mean, right now like my process
0:27:42 for YouTube thumbnails is pretty much created an image
0:27:47 with AI and then inject my face into it
0:27:49 and then pull it into Canva and add the text on top.
0:27:51 ‘Cause I haven’t really gotten the AI generators
0:27:55 to sort of stylize the text I like
0:27:58 or like put it in the exact location that I like yet.
0:28:02 So, if Idiogram, if you can go and get that granular,
0:28:04 I want the text left aligned
0:28:05 in the bottom left corner of the image.
0:28:09 And I want this, you know, this image aspect,
0:28:10 this the aspect of the thumbnail
0:28:12 to be up in the top right corner.
0:28:14 Like if you can sort of dial that stuff in,
0:28:17 that’ll be, that’ll be really, really awesome.
0:28:18 That would save a lot of time
0:28:20 and it’ll save a few extra steps for people.
0:28:21 – For sure.
0:28:22 – Very cool.
0:28:23 – Do you have any spicy ones
0:28:25 like on our episode the other day?
0:28:27 – Oh, how I make all of my AI girlfriends.
0:28:31 (all laughing)
0:28:34 My harem of AI girlfriends.
0:28:35 So, here’s what I’ll share.
0:28:39 So, this one is going to be a little bit more in the weeds.
0:28:43 This is actually how I add tools to future tools.
0:28:46 Now, this is probably not going to be super relevant
0:28:48 to everybody watching,
0:28:50 but I think this should give people ideas
0:28:53 of the types of workflows they can build
0:28:54 to sort of run their businesses
0:28:56 and make them a little bit more automated
0:28:58 behind the scenes, right?
0:29:01 Obviously, this exact step-by-step process
0:29:03 is not going to be something you’ll need in your business,
0:29:06 but hopefully it gets the wheels turning
0:29:09 of what these tools are capable of.
0:29:11 So, when I come across a new tool
0:29:13 that I want to add to future tools,
0:29:14 I actually came, there’s a tool
0:29:17 that I’m going to add to future tools
0:29:19 right now going through this process.
0:29:20 There’s this tool called Spotter,
0:29:23 which is a YouTube tool,
0:29:26 which helps you come up with titles and thumbnails
0:29:28 and break down hooks for your YouTube videos
0:29:29 and stuff like that.
0:29:31 It’s a tool that I use, right?
0:29:33 But it’s not on future tools yet.
0:29:35 So, when I want to add it to future tools,
0:29:36 here’s what I do.
0:29:39 I copy the URL in my browser.
0:29:42 I come over to this Google Sheet spreadsheet.
0:29:45 I have a little tab that I call the machine
0:29:49 and I plug in the URL here to this input URL.
0:29:53 Then I have a make.com automation setup
0:29:55 that does a whole bunch of steps.
0:29:58 So, all I do is I plug in this input URL.
0:30:00 I go to make.com.
0:30:02 I have, they call, like they’re,
0:30:04 you know, in Zapier, they call them zaps,
0:30:06 make.com, they call them scenarios.
0:30:10 So, I have this scenario called integration scraping bee here.
0:30:13 And what it does is it watches that Google Sheet
0:30:16 that I just showed you for a new entry
0:30:18 into this input URL here.
0:30:21 And then it uses a site called scraping bee,
0:30:25 which goes and takes all of the content
0:30:27 that’s on that website.
0:30:30 So, it will go and look at spotter.la
0:30:32 or whatever the URL was.
0:30:33 It will look at that site
0:30:36 and it will look at all of the text on that site
0:30:37 and it will scrape it all.
0:30:40 Once it scrapes all of that content,
0:30:43 it then goes into a chat GPT.
0:30:47 This is actually using GPT for right here.
0:30:50 It will actually go into GPT for,
0:30:53 and you can see my prompts that I give it here.
0:30:56 Summarize what this tool does in a single paragraph,
0:30:58 include what it can be used for
0:31:00 and why people might want to use it.
0:31:03 Your response should begin with the name of the tool.
0:31:07 Right, so basically like spotter does x, y and z.
0:31:09 And so, it takes everything that was scraped.
0:31:12 You can see the message content here is what was scraped.
0:31:16 And then additional message content is to summarize this.
0:31:19 So, it summarizes it into a paragraph.
0:31:21 And then it takes that paragraph
0:31:23 and it tells it to summarize it
0:31:25 into a short little snippet, right?
0:31:28 So, if you go to the Future Tools website,
0:31:30 the homepage shows just like a one sentence snippet
0:31:31 of what a tool does.
0:31:32 And when you click in,
0:31:36 you see a longer paragraph about what that tool does.
0:31:39 So, this second run through of chat GPT here,
0:31:42 it says in as few words as possible,
0:31:44 describe what this tool does in one sort sentence.
0:31:47 Your response should begin with the name of the tool.
0:31:50 So, same idea, it just takes the longer paragraph,
0:31:53 moves it to the next step in this workflow here
0:31:56 and makes a one sentence version of it.
0:32:00 And then the last step here,
0:32:03 so the last one basically just pulls in the tool name, right?
0:32:04 It just looks at the website,
0:32:06 says what is this tool called?
0:32:07 And it pulls in the tool name
0:32:10 by looking at the website here.
0:32:11 So, what is the name of this tool?
0:32:13 Your reply should have no extra sentence or details.
0:32:15 Please simply output the name of the tool
0:32:17 and nothing else.
0:32:19 And then once it does that,
0:32:21 it takes everything it just pulled,
0:32:23 it scrapes it, makes a long paragraph,
0:32:25 makes a short sentence, figures out the title,
0:32:27 and then injects it back into the Google sheet.
0:32:30 So, let’s pretend I’m pulling in otter.ai
0:32:32 and it’s not on the website yet.
0:32:35 I could put the input URL is otter.ai.
0:32:37 I’m gonna make sure it’s running in the right thing.
0:32:38 I’ll press run once
0:32:40 and now it’ll run through the whole process.
0:32:42 It’s scraping the sales page,
0:32:45 using chat GPT to summarize it.
0:32:47 So, we can see it pulled in the tool name
0:32:50 of otter.ai here over on the short description.
0:32:52 Otter.ai provides automated note taking
0:32:54 and real-time transcription for meetings.
0:32:55 Over in the long description,
0:32:59 you can see here that it’s a much more in-depth explanation
0:33:01 of what otter.ai does.
0:33:04 And so, now it’s basically created my short description,
0:33:07 the title, the long description,
0:33:10 and it also automatically pulled in
0:33:12 the featured image from the website.
0:33:13 That’s not an AI thing.
0:33:15 That’s just something that I have it import
0:33:16 through Google Sheets with.
0:33:18 And then once that’s done,
0:33:19 I actually have a separate little automation
0:33:21 that runs after that,
0:33:23 that basically creates the short link
0:33:25 and uploads it to Webflow for me.
0:33:27 So, that’s the whole workflow.
0:33:30 I told you it’s sort of in-depth and in the weeds,
0:33:32 but I wanted to show it because
0:33:35 I really love this make.com site, right?
0:33:36 If you’ve ever used AppEars,
0:33:38 AppEar can do a lot of the same kind of stuff,
0:33:41 but you can create these amazing workflows
0:33:43 where it does things like scrape a website
0:33:44 and then create a long paragraph,
0:33:47 create a short paragraph, grab the title for me,
0:33:49 and then put all of that information
0:33:50 into a spreadsheet for me.
0:33:53 So, whether you’re doing any sort of data analysis
0:33:55 and you’re going and trying to pull a bunch of data
0:33:57 off a whole bunch of websites,
0:34:00 you can use a workflow like this, right?
0:34:02 It lets you do a whole bunch of stuff in bulk.
0:34:04 And this will also work if,
0:34:06 let’s say I have like 20 tools listed here,
0:34:08 it’ll just one at a time go through the list
0:34:12 and follow that same process for every tool in the list.
0:34:14 – Yeah, I mean, that workflow is awesome.
0:34:15 I just, the whole time you’re talking there,
0:34:17 I’m sitting there thinking like, that’s really cool,
0:34:19 but it’s also very complicated.
0:34:21 And like, this is the kind of stuff
0:34:24 that agents are gonna like simplify, right?
0:34:25 Right, right.
0:34:26 And so, and I was wondering like,
0:34:28 okay, makes cool, is AppEars cool?
0:34:30 Do those sites still exist in the future?
0:34:31 Like when you have agents?
0:34:33 Like maybe that’s why Darmesh is making
0:34:34 that agent.ai thing.
0:34:38 Maybe he’s realized like those kind of sites
0:34:40 are gonna be simplified down into like,
0:34:41 okay, here are the agents.
0:34:42 And here’s like the kind of the instructions
0:34:44 for the agents.
0:34:46 And then you just hand the agents your API keys
0:34:49 and it does all that automatically.
0:34:51 – I mean, it’s a little bit more hands off
0:34:53 than I think I made it look
0:34:56 because I had to like turn off the automations
0:34:57 from running automatically
0:35:00 so that I could demo them on the video.
0:35:02 So literally what I’m doing when I add a new tool
0:35:07 to the website is I’m plugging in the URL of the website.
0:35:10 And then it just goes and does the rest, right?
0:35:12 It’s watching the Google sheet
0:35:15 for new information being added in like,
0:35:17 I don’t know, every half hour or every hour
0:35:18 or something like that.
0:35:19 It comes back, double checks the Google sheet
0:35:22 to see if there’s any new information on the Google sheet.
0:35:24 If there is, it goes through the list
0:35:25 and runs the automation.
0:35:29 So my role in all of this is literally just adding a URL
0:35:31 to the input URL on the Google sheet.
0:35:33 And then all of the rest of it happens
0:35:35 behind the scenes automatically.
0:35:38 So it’s sort of like I created my own little AI agent,
0:35:39 but you’re right.
0:35:42 I think in the future, I won’t have to build all that.
0:35:45 I’ll just have to tell like an AI or like a chat bot,
0:35:47 here’s what I want you to do.
0:35:49 And it will build all those automations
0:35:51 that I just sort of showed you the workflow for.
0:35:52 – Yeah, but then I’ll, like I said,
0:35:54 I’ll probably just ask you your API keys.
0:35:55 It’ll save that somewhere secure.
0:35:58 So it remembers it for the future
0:35:59 and it’ll just do all of that.
0:36:00 That’s gonna be so awesome.
0:36:01 – Yeah, yeah.
0:36:03 So share something less complex
0:36:06 ’cause that one was really in the weeds.
0:36:10 – Mine’s way less complex, so yeah.
0:36:14 So this is something I started doing recently.
0:36:16 So right now I’m kind of working,
0:36:17 I’ve had lore.com for a long time
0:36:18 and I’ve been trying to figure out what to do with it.
0:36:20 I’m still kind of trying to figure out
0:36:21 what to do with it beyond.
0:36:23 Like I like writing my newsletter,
0:36:24 but I feel like it’s such a great domain.
0:36:26 I should be doing something more with it.
0:36:29 And originally I bought the domain
0:36:31 ’cause I was trying to do the movie studio with Barry Osborn.
0:36:33 And so I’ve been interested in using it
0:36:35 for something more entertainment related for a long time.
0:36:38 So right now I’m doing this thing called Lore Labs.
0:36:40 I’m like, it’s kind of like a AI video agency
0:36:42 and a kind of community to help people
0:36:44 figure out how to create AI videos as well.
0:36:45 Still figuring it out.
0:36:46 So I’ve been working with these two great guys,
0:36:48 one of them who makes some of the best AI videos
0:36:49 I’ve ever seen.
0:36:50 I’m not announcing who he is yet,
0:36:52 but we’re collaborating on it.
0:36:56 And so we’ve got a slack full of different conversations
0:36:59 about like, okay, what should be our strategy, pricing?
0:37:01 What’s the initial strategy?
0:37:02 What are we gonna try if that doesn’t work?
0:37:04 Like, you know, lots and lots of conversations
0:37:06 to figure everything out.
0:37:08 And then we’re having weekly meetings.
0:37:10 And then one thing I realized that AI is really good at,
0:37:12 and I think other people could use it this way too,
0:37:14 is like using AI is almost like a meeting agenda.
0:37:16 Like, okay, you’re having a meeting,
0:37:18 you’ve been having all these conversations
0:37:20 through email or Slack or whatever.
0:37:23 Just throw them all in there into Claude
0:37:26 and you make a custom project for it or whatever.
0:37:29 And they say, hey, what are the actual like things
0:37:30 that need to be decided upon?
0:37:33 What are the like action items you need to do?
0:37:35 You know, key takeaways from the conversations.
0:37:37 And then just have it create the meeting agenda
0:37:39 automatically that way versus like,
0:37:40 well, you’re gonna sit there like 30 minutes
0:37:42 and like type up the meeting agenda
0:37:44 and have to re-remember what you actually talked about.
0:37:45 – Yeah, yeah, yeah.
0:37:46 – So that’s what I’m doing now
0:37:48 is I literally just copy and paste all of it in there.
0:37:50 And like I’ve told it, here’s how I want my meetings
0:37:53 structured, here’s the conversations,
0:37:56 figure out the agenda, and then send it out.
0:37:57 – That’s super smart.
0:37:59 So you’re basically, let’s say you’ve got some emails,
0:38:03 you’ve got some DMs, you’ve got some tweets,
0:38:06 I don’t know, you’ve got all of these various places
0:38:07 that you’ve had communication.
0:38:10 You just pull all of these pieces of content
0:38:13 in these communications you’ve had,
0:38:14 put them into Claude and say,
0:38:16 I’m about to meet with this person
0:38:20 based on everything I just put into your input.
0:38:22 How should we like, how should we keep
0:38:25 this meeting on track, right, an agenda for me?
0:38:26 – Yeah, totally.
0:38:28 And ideally there’s, you have like an agent
0:38:29 for that in the future, right?
0:38:30 – It actually goes out there and like,
0:38:32 yeah, we’re talking on email, we’re talking on Slack,
0:38:35 and you know, whatever, yeah, go and get that for us
0:38:36 and then actually create the invite.
0:38:37 That’s like the next step, right?
0:38:39 To actually do all of it.
0:38:41 But right now, yeah, I’m manually copying and pasting
0:38:44 and then just, it still saves me probably 20 minutes or so.
0:38:45 – Yeah.
0:38:46 – But yeah, ’cause some meetings it’s like,
0:38:48 okay, what did we talk about in the past?
0:38:50 Or what’s actually important to discuss?
0:38:52 You know, just tell the AI like what you want.
0:38:53 Like what’s, you know, how do you like meetings
0:38:55 to be structured?
0:38:55 – Yeah, well what’s–
0:38:58 – Feed the day to day and it’s pretty, it’s pretty good.
0:39:00 What’s even cooler too now is this is just
0:39:01 a brand new feature.
0:39:03 The week that we were recording that there was a new feature
0:39:05 in Google Meet where now we’ll summarize the meetings, right?
0:39:07 This has been in Zoom for a little while now,
0:39:10 but Google Meet just rolled out the feature where it will,
0:39:12 if you turn it on, it’s not always listening to your calls,
0:39:14 but if you turn it on, it will listen to your calls
0:39:16 and basically summarize the meeting for you
0:39:20 and make like a to-do list off of the meeting,
0:39:22 like an action item list kind of thing.
0:39:24 And so let’s say you need to have a second call
0:39:26 with that person, a follow-up call.
0:39:29 Well on that follow-up call, you just grab the meeting notes
0:39:32 from your last call, plug that in and say,
0:39:34 here’s what we talked about last time, you know?
0:39:36 – Yeah, yeah, yeah, what progress has been made,
0:39:40 what’s unresolved, you know, what do we need to discuss?
0:39:40 Yeah, it’s awesome.
0:39:42 – So really, really cool.
0:39:45 No, I think these are some awesome ideas
0:39:47 and I really like this format.
0:39:50 Like I love this format of just here’s some
0:39:51 of the cool ways we’re using them.
0:39:53 I think we’re gonna get to a point where we’re like,
0:39:54 all right, we’re running out of use cases
0:39:56 that we use ourselves,
0:39:58 but we’re constantly out there like fishing
0:40:00 for other use cases, we’re watching Twitter
0:40:03 and YouTube and Reddit and all these places
0:40:05 where people are sharing their cool AI use cases.
0:40:07 So I think it’ll be cool.
0:40:09 I think maybe next time we do one of these,
0:40:10 we’ll bring like a bunch of like,
0:40:13 here’s some cool workflows that we came across
0:40:14 that other people are using
0:40:15 that we think you should know about.
0:40:17 No, but this has been a fun discussion.
0:40:20 Once again, let us know your thoughts in the comments
0:40:21 wherever you’re listening to this.
0:40:23 If you’re on YouTube, let us know your thoughts.
0:40:25 If you have some cool use cases that you wanna share
0:40:27 and you want us to maybe shout ’em out on a future episode,
0:40:28 share them in the comments.
0:40:32 That might be a really great way to discover new workflows.
0:40:33 If we talk about your workflow,
0:40:35 we’ll shout out your name on the episode
0:40:38 and let everybody know that that came from you.
0:40:40 But I think we’re probably gonna do more of these episodes.
0:40:42 So let us know what you think of them in the comments
0:40:44 either on YouTube or Spotify
0:40:46 or wherever you’re listening to podcasts.
0:40:48 And if you wanna make sure more of these podcasts
0:40:50 show up in your feed, make sure you like and subscribe.
0:40:51 We really, really appreciate it.
0:40:53 It makes us feel a little warm and fuzzy.
0:40:55 And thank you so much for tuning in.
0:40:57 – Yeah, thank you so much.
0:40:59 (upbeat music)
0:41:02 (upbeat music)
0:41:05 (upbeat music)
0:41:07 (upbeat music)
0:41:10 (upbeat music)
0:41:12 you
0:41:14 you
Episode 23: How can AI simplify complex workflows and enrich language learning? Matt Wolfe (https://x.com/mreflow)) and Nathan Lands (https://x.com/NathanLands) take you on an insightful journey exploring diverse AI use cases and tools.
In this episode, Matt and Nathan delve into the intricacies of learning Japanese and coding with AI, leveraging tools like Claude and Perplexity to streamline and enhance these processes. Nathan shares his experience simplifying Japanese language learning through targeted translation techniques, while Matt reveals his tips for efficient coding using AI, along with strategies for optimizing content with AI tools like Perplexity and Ideogram. The duo also discusses workflow automation, potential SEO hacks, and meeting management with AI, rounding out the episode with engaging and valuable insights.
Check out The Next Wave YouTube Channel if you want to see Matt and Nathan on screen: https://lnk.to/thenextwavepd
—
Show Notes:
- (00:00) Exploring advanced AI use cases for businesses.
- (04:40) AI recreates Doom from gameplay videos accurately.
- (07:30) Useful for newsletters, business documents, and data management.
- (09:55) Custom instructions enhance model quality and results.
- (13:09) Create shorthands for tasks using custom instructions.
- (18:19) Perplexity’s page feature generates mini Wikipedia entries.
- (25:25) Curious about AI-generated YouTube thumbnail prompts.
- (26:48) Automating business workflows.
- (31:49) Automated workflows simplify data analysis with make.com.
- (35:31) AI efficiently simplifies meeting agenda creation process.
- (37:21) Google Meet now summarizes and generates meeting notes.
—
Mentions:
- Riley Brown: https://www.youtube.com/channel/UCMcoud_ZW7cfxeIugBflSBw
- Ideogram 2.0: https://about.ideogram.ai/2.0
- Perplexity: https://www.perplexity.ai/
- Claude: https://claude.ai/
- Make.com: https://www.make.com/en
—
Check Out Matt’s Stuff:
• Future Tools – https://futuretools.beehiiv.com/
• Blog – https://www.mattwolfe.com/
• YouTube- https://www.youtube.com/@mreflow
—
Check Out Nathan’s Stuff:
- Newsletter: https://news.lore.com/
- Blog – https://lore.com/
The Next Wave is a HubSpot Original Podcast // Brought to you by The HubSpot Podcast Network // Production by Darren Clarke // Editing by Ezra Bakker Trupiano