He Built An AI App That Lets You Talk To Podcasts

0:00Welcome back to the Build in Public Podcast, and in this episode, I brought Paul back on to talk about his new app he just launched called PodSips, which allows you to communicate with all your favorite podcasts. We also break down how he built it and his future for monetizing that app, also breaking down the future of AI, what Open Claw's doing, what Anthropic is doing. We'll see in the episode. Paul, welcome to the podcast. You've been on a few times. I want to get to know what you've been building because you launched something pretty amazing recently. First of all, before we get into it, like, you know, back up to just a little brief background of, you know, what you built and who you are, and we can deep dive later down the road later in the podcast, but just give us like a 30,000 point of view foot point of view of like

0:46what you've built and and a little bit about your background. >> Yeah, so I've been involved with and playing around with and building businesses with technology since 2016. I built and then sold with a business partner of mine technology startup, and then in that business, we were heavy users of Chat GPT and the various GPT APIs, and I fell in love with AI. And I've been traveling a lot over the last 4 years, and because I don't stay in one place, I listen to a lot of podcasts, and my podcasts are like my friends. Like, I don't have friends to go get coffee with because I don't know anybody in each of the places that I travel to. So, the the podcasts have

1:32been my family. The the podcasts are the the people that I talk to, and I have just become utterly obsessed with podcasts. So, I decided to merge those two worlds and create an app called PodSips. And I created it because I Yeah, there it is. I created it because I would be out on a walk or driving or out on a run or whatever, and I'd be listening to a podcast, and I like technology, I like science, I like things that are sometimes difficult to understand. I like like deep finance, and I have found that or I did find that it was hard to understand the things that I was listening to from the most popular podcasts in the world even. So,

2:19I took it upon myself to build a podcast app. It's on the iOS store. You can download it now. It's called PodSips, and it allows you to listen to a podcast and effectively have an AI in the room with you listening to that podcast with you, so that you can say, "Hold on a second. I don't understand what they just said. Explain that to me." And it will say it it'll explain it back to you what was just said without you having to say, "Here's the whole transcript. Here's all of the stuff. I'm currently at 15 minutes and 38 seconds." Whatever. But then, to make all of that possible, I found that all of the technology to make all of that possible also afforded me the ability to chat with not just one podcast episode, not just one entire

3:06podcast series, but chat with entire podcast series together. So, with PodSips, you can effectively communicate with any number of podcast episodes and any number of podcast series all at the same time while also being able to listen to a podcast and communicate with an AI that's in the room with you listening to the podcast. So, it's an AI-enabled podcast platform that I built to solve the need in my own life because my podcast friends just weren't making sense. Yeah. Yeah, I love that. So, how does it work like under the hood? Like you said you could chat with multiple podcast episodes, and it what like what does that work look like as far as like

3:51per user they can have a specific, you know, niche of podcast, and it pulls the information that's relevant to their question, you know, based on >> Yeah. their podcast that's preloaded or has queried the path. >> yeah, that that's that's a good way to put it. But just to get a little bit more granular, I'm I'm very interested in health and let's say running. So, if I got one of the podcasts that I listen to all the time is the Rich Roll podcast. So, if I listen to the Rich Roll podcast and he has some ultra runner on that's talking about how to fuel at mile 50 in an ultra marathon, but I know that he also talked to somebody else that maybe was a doctor or whatever that was saying something adjacent to this. What I can do with PodSips is I can load the entire

4:38podcast series, every single episode from the podcast series, and say, "What is the right way to fuel at mile 50 of an ultra marathon according to all of the episodes from this entire podcast series?" And then if I wanted to add another podcast series on top of that and say, "What do all of the episodes from these 15 podcast series say about fueling during an ultra marathon or creatine intake or the right way to taper before a long run whatever it might be?" All of that is possible because of the architecture that I've built behind the scenes to ingest all of those podcasts and make them effectively accessible to the AI.

5:23So, yeah, that that's how PodSips works from a high level anyway. Yeah, that's awesome. And who is your ideal like avatar? Like, you know, who's it for? Like I could see myself spending hours in this because I listen to other podcasts, but then I forget what they say or you know, I have to go back. When when did they say that or when did they talk about that specific product? So, like am I your ideal customer or is like who is it built for specifically? >> Yeah, it's built specifically for the person that understands that podcasts are a medium of trans a medium to transmit information Okay. like news or intelligence or learn things. You know, there are so many podcasts that are hilarious comedic podcasts like Matt and Shane's Secret

6:08Podcast is one of my favorite podcasts to just laugh to, but you might not have questions about whatever Shane Gillis and Matt McCusker are saying on Matt and Shane's Secret Podcast, but you might have questions about what Joe Rogan talked to this scientist about and how that compares to what this other scientist said. So, I'd say the ideal customer, the ideal user is somebody who enjoys learning from podcasts and wants to unlock the knowledge that is in podcasts by taking it and kind of changing it to make it more approachable for their own ears. Yeah. Yeah. Yeah, what do what would you say is like the hero feature of the product? Like what's the main, you know, product that people would or the main feature of

6:53the product that people would subscribe to and go like, "This is insane. This is amazing." >> Yeah, while you're listening to any podcast episode, the AI is aware of what you're currently listening to. So, at any point while you're listening to the podcast, you can tap a button that says, "Explain this." And it will automatically come back to you with a message that says, "Here's what I don't know. Here's what Chamath just said on the All-In Podcast about some technical financial thing or his outlook on the economy because of AI." And it will explain it in layman's terms very simply, and it takes, you know, all of three or five seconds. So, you can pause the podcast, read that, takes three to five seconds, and then go back, and then have a significantly better listening

7:39experience because you're like you're on the same playing field. You're on the same level now as the person who's talking. That's what I'd say the hero feature is. I love that. Think if we had that for books, you know, it's like there's many times where I'll be reading a book. I'm like, "Man, do I have time to go look this up on my" You know, you're reading a book and then you have your laptop on your left hand, and you're like, you know, "What is what does that mean?" That's important part of the learning process is you could read, but if you don't understand the concepts, what's the point? You know, so same with podcasts is like I could listen, but if I don't know what they're saying, we'll have a misconception, and so that solves that problem. And yeah, you're right. Podcasts are so many different types. Still, the category in itself is growing massively, but as far as education, health, finance, things like that, that could be

8:26a huge unlock. And and you had just released this after how long working on it? You know, wasn't >> Yeah. Yeah. Yeah, exactly. So, it's funny that you bring up books because and this ties into how long I've been working on this. So, when we were finished with our last company, I at the as we were like winding down that company and getting ready to sell it, I was listening to Alex Hormozi talk on I think the 100 Million Dollar Offers book, and I was out on a run listening to it, and I was like, "I love Chat GPT, but I can't understand [snorts] what this guy Alex Hormozi is saying, and I don't know how to take what he's saying and apply it to my business." And I was like, "Oh, it would be so cool if

9:11I could use the intelligence of Chat GPT, give it the transcript of the book, and say, 'Explain this to me given everything you know about me and my business.'" And that was back in like 2023 maybe. I think the beginning of 2023. Anyway, and then I went down the path of how do I build PodSips but for audiobooks and for regular books. And I quickly found out that there's a humongous hurdle in the world of copyright and being able to take a book, ingest it, put it through AI. And I was like, "I do not want to get into copyright questions." I I actually posted in the some Reddit thread where there were a

9:56bunch of authors, and I was like, "Hey, I think I'm going to make this app. What do you guys think about this app?" And I got completely roasted. Like just completely destroyed by the authors in there like, "I do not want this. My readers do not want this. Everybody hates AI in this world." And I was like, "Okay, that's a sign. I'm not going to do this for books." That was 2023. "I'm not going to do this for books. I'll put this on the back burner." Sold that company and then I decided I'm going to do it for PodSips. I'm going to do it for podcasts and create PodSips. So, that's a long way of me saying the idea has been in there for a long time, but I only started working on it, quote-unquote, only started working on it in mid-2024 after the last company and I've been working on it on the side

10:41since then. So, it's been a long time coming, to say the least. Yeah, and it's cool to see something come alive, no matter even if it takes years. It's something that you love. It's something that you've stuck with and it solves a problem, too, in the day of uh distractions and infinite information out there. It helps you organize your information and your thoughts. And I think it just helps you learn more efficiently. Um you know, everyone learns differently. For me, I I like to learn slowly and understand kind of what's happening before I could start to apply it. Um and I think that for anyone that has like a little bit of an ADHD, um I mean, that's a perfect combination, you know? Do you know of any other app or any

11:26other product that does something similar? I mean, it could be a good thing to have competition, but I can't think of anything else. Yeah, there are a couple apps that actually got to this world in the podcast space before I did. My understanding of what they offer, though, is they offer the listening companion, similar to how PodSips offers the listening companion. But, my understanding is that no one, except for PodSips, offers the ability for you to chat with multiple podcast episodes at the same time or entire series at the same time and all of the content of all of those episodes or multiple series at the same time. Like right now, if you wanted to, you could go chat with all, like, I don't know, 800,000 episodes of

12:13podcasts on PodSips and ask questions. It's going to be an odd answer you're going to get back because all of the podcasts are about so many different things, but nonetheless, it's possible. So, yeah, there are some companies, um I think, that are in the space that are adjacent to what I'm doing, but I'm a I'm very much heads-down trying to make PodSips work. Yeah, absolutely. And for anyone that's listening and that are interested, I'll put the link down below. You should try it out. >> Thanks, Bo. Yeah, yeah, absolutely. Absolutely. So, like the stack to build this, you know, what a lot of people are watching that are builders in on this channel. And you know, what what went behind it? Like is it intense infrastructure? What did you use to build it? How'd you do it? >> Yep. Yep. So, I I'm probably going to I'm going to talk too much. You can You can cut me

12:59off and cut straight to the answer if you want. >> [laughter] >> Um Claude code, that's it. There's nothing else. Yeah, sorry. Yeah, secret Claude code. Just ask Claude. Um any So, in early 2023, right after ChatGPT came out, I was actually talking to ChatGPT trying to understand the idea of vector embeddings and what does it mean to embed a piece of text and how do embedding models work and what is vector space, all of that stuff. And that was the impetus that really pushed me across the edge to say eventually, whenever I was ready to build PodSips after the last company, that's the thing that pushed me over the edge to say, "Oh, I

13:44understand how I can build this." So, the tech stack is a pipeline that ingests podcasts and their transcripts, chunks, embeds, stores them in a vector database, and then makes that available, that vector database available, um searchable, um and filterable by uh by an AI. So, I uh I use Railway for the back end. Um the app is built on Expo and React Native. It's been fantastic to work with Expo and React Native. Um I use Claude code for absolutely everything. Um I've actually it's a middle middle to end of April 2026 and I've actually transitioned away from using OpenClaude and use Claude code

14:31instead of OpenClaude because Claude code is just so darn good. Um Claude code runs my life. Claude code built PodSips um and it's been great. So, vector databases, Railway, uh TypeScript, React Native, um Expo, and Claude code. Lots and lots and lots of Claude code. Cool. So, by doing the vector embeddings, that is how you're able to store so much data. That's right. Yeah, and make it semantically searchable. Exactly. Mhm. Cool. That's awesome. So, yeah, I'm just trying to wrap my mind around it. I mean, what's like reception been like? You know, you just launched it officially a month ago. Have you had any feedback or like what are people saying? Yeah, so I actually two days ago had a call with a

15:17user. Um I just got some feedback. He said, "I want to give you some feedback." And I was like, "Oh my goodness, yes, I would love that. Um please give me any feedback that you have." And he and I sat down for about an hour and 15 minutes and he said, "Before I give you any negative feedback, I just want to tell you what an incredibly magical experience this is. This is something that I didn't know that I needed until somebody opened this door and said this is a new way to consume media." And that's just music to my ears. Um so, the there have been a couple people that have said that to me. Um I haven't really pushed it out there and really got PodSips out there because I'm an active user of it and I keep finding little things that I want to improve. I'm not creating new features. I'm just like making little improvements

16:02on how fast things load and how different user interactions and user flows are in the app. Um but, the reception so far from the handful of users that I've put it out to has been really, really great. Really, really well received. That's cool. I love that. Yeah, the best feedback loop is from your best customers, right? And and I I I always think like going back to when I first built Easy Flip, you know, how perfect I wanted to make it. I'm like, you know, I can't have this button not work or like it works on my machine, but it doesn't work on yours. Oh, okay. Well, you know, and and I just knew that it gets to a certain point where you just got to release it. You know, like cuz they're going to find bugs that you can't reproduce on your end. Um so, yeah, creating that feedback loop and then

16:48keeping those lifelong customers, you know, making sure like, "Hey, I'm going to do whatever it takes to make sure that this app serves you and performs what it was intended for." I like I feel like that's not talked about enough. It's like we we talk about minimizing churn. Well, it's like talk to your users. Figure out what's making them churn and solve the problem, you know? Uh and that's a great way to to build out something maybe you have in your mind like, "Oh, I wanted to do this." But maybe they come up with something great that you hadn't thought of before. Um like what's what's the next step for PodSips? Like what is what does monetization look like? What does marketing look like for you? What's the next step for you and your project? >> Yeah, it's a great question, Bo, and it's actually something that I wanted to talk to you about. We can talk about it

17:33on here. We can talk about it offline. But anyway, it's it's something that I wanted to talk to you about. Um I've had an interesting situation come up where a a couple podcasters have reached out to me, people with sizable podcasts, and they have said, effectively, something something like, "Hey, I am publishing my podcasts, but the platform that is my publisher doesn't give me the ability to hand my podcast data to an AI agent." And you and I have talked about this extensively. I think that any company that doesn't make an AI agent a first-class user of the product is going to be replaced by somebody else. So, for

18:18example, with PodSips, you can go to developer.podsips.com and you can get the OpenClaude skill and the Claude code skill and effectively hand the PodSips database and the ability to search the PodSips database, you can hand that to your AI. So, you can imagine like you're talking to your OpenClaude and you're like, "I know that somebody talked about this on the Andrew Huberman podcast. Go use PodSips to go figure out what they said." Your OpenClaude will have the ability to use PodSips and the data inside PodSips without you having to use the mobile app. So, it's another it's another surface for people, in this case, AIs,

19:04to access and utilize the value of PodSips, which is that gigantic vector database. So, this all comes back to what I was saying about the podcasters reaching out to me and saying, "Our platforms that we pay for don't give us the ability to tap into an API or tap into a CLI or tap into an MCP that tells us or gives us the ability to give our agents our podcast data." So, you can't, as a podcaster, say, "How did my last three episodes perform against each other?" Or "What was my podcast performance like um across YouTube, Spotify, and the RSS feeds in the last 90 days?" You effectively can't do that with these podcast apps. So, the idea

19:50has come to me through these podcasters that have said, "What if PodSips, the user customer-facing app, was the way that people interacted with the podcast?" But then the back-end, the PodSips for creators, if you will, aggregates all of that data, YouTube, Spotify, RSS feed, aggregates all of that data, and then makes it accessible to the podcaster, so that the podcaster can say, "Go use PodSips. It's a great app. It affords you this new ability to interact with my content." And then they can use PodSips for creators on the other side and have access to all of their data, PodSips data, RSS feed data, YouTube data, and Spotify data all in one place and hand

20:37that to their AI. So that PodSips for creators, I think is going to be my main way of marketing this instead of going direct to consumer. That was a super long answer, but I hope that it answered your question. Yeah, cuz I was wondering if it was B2C or, you know, direct to, you know, big business. And I mean, that all makes sense. So basically you would make money off of what, the API calls or like So the app is still paid. You you still have to pay for a subscription to use the PodSips podcast app because there is an expense for serving the API queries and getting all of the context from all these podcasts. Um It's for fun. >> [laughter] >> Yeah, yeah. No, I get it. You have you have to I mean, I'm sure, you know, each one adds

21:24up. I don't know what model you're using specifically, but even if it's a cheaper model, it's still going to add up. Yeah. So yeah, I didn't even think of that. That is really cool. So essentially now you can go to big, you know, podcasters and say, "Hey, do you want all your podcasts to be, you know, you can query it at any time about anything?" Um yeah, that's pretty cool. Wow. Yeah, you can query your podcasts to see to ask any questions, and so can your listeners. But also, what if I gave you the ability to query your data and talk to your AI about your podcast data? And in exchange for offering that, Mr. Podcaster, can you promote PodSips, the actual mobile app? Perfect. Oh my. I mean, that is that is the the marketing strategy,

22:11right? I mean I sure hope so. Yeah, yeah, absolutely. Um especially if you have people already reaching out, you know, that's kind of your validation right there. Yep. Yeah, that is a pretty cool. Wow. >> You can be a you and the Building Public podcast can be a a tester of PodSips for creators, Beau. So essentially I download that skill, give it to my Openclaw, and now anytime when I'm building with Openclaw, I can always use that as a reference. You know, I can even keep that my Building Public agent, you know, just specifically for her. Wow, that's pretty cool. >> Yep. So two two separate There will be two separate skills. So there will be the skill that is number one, which is you can interact with the podcast information, like the the transcripts of the podcast, the content of the podcast.

22:56That's what you and anyone else can find today at developer.pod sips.com. There's an API and a skill that hands the podcast content to your AI agent. Okay. But then the second thing that I'm talking about building is data aggregation across your YouTube, where you publish your podcast, your podcast, where you publish your pod your your Spotify, where you publish your podcast, and your RSS feed, and PodSips, where you publish your podcast. Taking that and giving you another skill, another API that you can hand to your agent and say, "Tell me about my podcast data. Tell me what other podcasts in my niche are talking about this week that I can talk about. Tell me what guests they're

23:42having on that I can potentially court as a guest for my podcast." So there's two separate sides, PodSips and then PodSips for creators. So PodSips for creators is in the works, and that's my hopefully my primary marketing strategy. There's a product I use for YouTube just like that explaining the same thing. It's called Tube Lab. And when my agency uses it to find outlier videos of other people in the AI niche, this is essentially built for podcasts. If that's how I'm understanding it. So yeah, I mean, that's pretty crazy. And I mean, that's the cool part. You build a cool, valuable product that's, you know, not only useful, but you loved building it. I mean, it should sell itself. Um and I love it when I

24:27you know, you know, you you build the product and the people will come. Um so that's very cool. Um well, is that what you're mainly thing you're building now? Or I'm I'm sure you're working on something else. Or is there anything else you're you're building and working on? Yeah, so I The other thing that I'm working on that I'm sure you and I could go deep deep down the rabbit hole on is I built out a complete Openclaw setup, and then I refused to pay the API costs from Anthropic. I just said, "I I I can't do this. I'm not going to pay multiple thousands of dollars per month for Openclaw to go check my email for me." >> [laughter] >> Um so I didn't really dawn on me that what Anthropic was trying to get people

25:12to do is move from Openclaw to using Co-work and using Claude Code instead of Openclaw. Because I had been using Claude Code for coding and building different apps, like PodSips and working with clients for a year and a half or two years. But then it like all clicked one day as my Openclaw was turned off, and I was like, "Oh, I can just move everything over into markdown files inside of Claude Code, skill files inside of Claude Code, and memory files inside of Claude Code." So the other thing that I'm actively working on is, unfortunately, because I love open source and I love Openclaw, but I'm working on moving my entire Openclaw setup over into Claude

25:58Code so that I can use Opus 4.7 for my agent. And unfortunately, it is so good. It is so good. >> [laughter] >> Claude Code is amazing, you know, and and unfortunately, the news yesterday, you can now switch it back if you want. >> I know. I know. You've already made progress. But that that's the same thing, too, is when they made that change a few weeks ago, I'm like, "Well, how can I get the best of both worlds?" Because you're right, telling Openclaw to check my email would cost $10 now, and it's like, "Well, that would be cool." Um so I had to, and like other people, um just get Claude Code open in your Openclaw folder, and then anytime I wanted to change a cron job, fix something, change a prompt, build out Mission Control, make the UI better, um

26:44I would just go to Claude Code. But it would understand all of the context relevant, you know, with my Openclaw setup. Yeah, I'm still doing both, but now that they officially came out and said, "Hey, you know, Claude CLI good to go." Yep. Then I don't need to pay, you know, because I was at about $700 for the month so far using um tokens. Wow. Switched it yesterday, and and I have it split with OpenAI. All the cron jobs are through OpenAI through their subscription. And so it's like a best of both worlds, but I still have Claude Code there, and I'm like, it's just so much of a better experience. And and now with Opus 4.7, you can really dial in the reasoning and the difficulty, the effort, I mean, Yep. you know, low and and max. Um it's those little things. And you can

27:29tell that Anthropic is directly the direct competition. Anytime Openclaw comes out with something, oh, Anthropic shipped three different new things that are better, you know. Um one thing that I've been playing around with was Claude Design. I don't know if you've checked it out, but cuz there's been so much released lately, but I've been loving Claude Design. Have you looked at it at all or I have. it? It's just another thing that's out, right? I I have a gripe with you, Beau, because you texted me and you were like, "Claude Design is great. It's awesome. It's so good." Dude, I I tried Claude Design because the whole point of Claude Design is that you can take a repo, you can take a folder full of code, you can upload it into Claude Design, and you can say, in my case, make this button and make

28:15this feature. Mock it up. Don't write the code that actually does the back-end stuff. It's not going to do that. But like make this visual feature inside of my app. I spent two and a half hours trying to get Claude Design to take my repo and my global styles and make this button inside of my app and make the user flow work with just mock data. Yeah. Beau, it did not work. I was very upset with Claude Design. For your mobile app, right? That's right. Yeah, I mean, try with our web app. You're saying it's a skill issue. Yeah. Yeah. No, it's not it's not. I hadn't I hadn't done anything with mobile, but I've just been in our community having people come in the call, we look at their landing page, and some one guy said his landing page was

29:01built in 2007. I said, "Well, we're going to fix that." I've been having fun with Claude Design, and we totally recreated his whole um basically it's just an intake form for leads Yep. in the Texas area, and he was completely blown away. So I think for that use case, if you have like an older landing page and you just need it to look modern and look better, um I think it's good for that because my prompts have been simple. I never I never hooked up my repo or anything. Literally just went to easyflip.com and said, "I hate my landing page. Recreate it in a different style, different variety." I like your landing page. And and yeah, we fixed it. So now it looks a lot better. But um yeah, so I don't know. It just depends. I think It does. It does. It does for sure. >> For landing page, it was great cuz just

29:47a few minutes we recreated a lot of people's landing pages. The cool part is is you can share that zip file or share right to Claude code and then it can build out the back end or plug it right into your repo, um you know, if you're if you use Claude code. Yeah, I think I think something just to touch on it really quick, it might be interesting for anybody listening. I think what we're talking about here, but we're not saying it specifically, so I'll try to say it specifically. I think that this is a common trend in this entire world of AI, where my impression of Claude design is like, "Ugh, it doesn't work." But your impression of Claude design is like, "Oh man, this is great." And I truly think that it's all in how you approach it. I think it's the the truth for all of these AI tools is like, "If I didn't come in, I knew that I was coming

30:33in swinging for the fences with my test, you know? I'm not I knew that if I went in and I was like, 'Let me try to just change this you know, the the color of a button on this website or whatever' or redesign a landing page.' I was very confident in Claude's ability to do that. But I was like, 'All right, let's put this thing to the test as if I had just spent two days, three days messing around with like the higher level stuff. Let's see if they can actually abstract out or extract out my uh global styles from my app and actually build this button." So, I think it's all in how you approach it. I'm sure that it's a skill issue, honestly, that I could probably get better with Claude design, but I was not happy with it, Bo.

31:18Um here, let me uh move I'll just show like some examples just real quick. So, for the audience can see. Like for me, it was just building out simple really like this this is a I was using Remotion for a lot of this, but I just basically gave it the magic hand uh link and I said create a video based on all this content on here. And in the first shot, it created this uh little video. Um Remotion, you know, takes a little bit more finesse, you got to go back with it, but it did a great job on the first shot. So, for like product um demos and things like that, I think it can helpful for people that, you know, don't want to spend time dialing in their Remotion skill. Yeah. It did a good >> is great, Bo. Um and then for Easy Flip, um I completely redid I took the Post

32:03Hog aesthetic and I Yeah, I was going to say. kind of like Post Hog and uh with all the colors and um did did a good job, and better than what it was. Um they can try the product out here uh by dragging Yeah, I love it, Bo. and and uh create their listing. So, th- then this took 5 minutes. That took the the video took 5 minutes, this took maybe 5-6 minutes. Um I created a an AI coach, that's a whole different product. Um and then for Content Mix, same thing. Um but yeah. You know, your your Easy Flip page made me think of something that we could probably talk for 3 hours on, but one of the most recent updates that I made, and I tried I tried to use Claude design for it, but I I got too angry. Um one of the

32:48most recent updates that I made to Pod Sips was whenever you onboard into the app, you can actually use the product before you even sign up and pay for anything. Um and your your landing page there for Easy Flip, I was like, "Oh, Bo's doing the same thing. Maybe you're just doing the thing that everyone teaches and I'm just a a complete noob here and it's it's how things should be done." But I um I have noticed that people are much happier and much much more likely to pay if they get to actually use the primary hero feature of the app, which is tell me what they're talking about um prior to being asked to pay. Yeah. Uh Claude suggested it. Like I just want to I don't have a free trial um just cuz I know it's I've gone back and forth with it and at least

33:33giving them to do one free listing, I'm like, "Okay, that's good." And then they could see Yeah. how it works. Like you drop the photos in, you click a button, and you get a title, pricing, item specifics, all you got to click is publish and it's done. Yeah. If you don't come from the eBay resell world, all of that you have to do manually. Go back to the photos, plug in what category am I putting this in, go fill all of the 25 item specifics according to what that item is, price the item. So, essentially that cuts all that out. Um Yeah. So, that's that's I was like, "Okay, I'll I'll I'll do that cuz the cost is also very low." Um a lot of these things we build have very good margins, but um that's I can afford one listing, you know, per perspective client. Yeah. But yeah, um going back to it, like there's so much keep hard to

34:19keep up with. I just kind of spent the most time this last week in Claude design helping other people out, too, but redesigning my stuff, too, cuz it just was so old. We did a a med spa for a guy in Georgia. We did this one for a guy who built a He's actually in Startup Empire um and he Yeah, he got cash from Elon. Yeah, from Elon and uh I was like, "Let's just see what it does." And in couple minutes, it uh did this for um you know, kids to help manage their uh allowances and stuff. But yeah, we we live in a new world. I I don't have the best um design skills, and I think that with AI out of the box, um to be able to at least create a little bit of a moat to be able to stand out in a overcrowded world nowadays, like, you know, how easy

35:04it is to just fire open Claude or Google AI Studio and and go from zero to one. I mean, you can go faster than ever. In 10 minutes, you can get 80% of your MVP work, and that's kind of scary. So, how are we going to stand out? I mean, design's one of them. And don't even get me started on the new GPT 2.0 I know. generation. I know. I almost texted you yesterday and was like, "Bo, are you putting this into Magic Hand or what?" >> [laughter] >> The thing with Magic Hand is everything's working, so I'm like, "Ugh." You know, you we always want to tinker around and like, "What am I going to break next?" So. Yeah, that's right. That's right. Yeah. Yeah, any other cool tools you've been working with? Have you tried Hermes agent? Uh you know, I I know you're a Cl- Do you have your Claude code tattoo yet? I mean, what what are you What are you using? Do you use anything else? Um I've been tempted

35:50to use Hermes for outreach. Yep. So, I've actually been poking around Grok and really using Grok inside of X, but I think the the biggest thing that is on the top of my mind for AI right now is and I don't know this for sure, but there's speculation that Mythos Mythos, whatever, from Anthropic and Spud, probably GPT 5.5, whatever, that these are the first models that we're getting access to that are trained on the I believe trained on the Grace Blackwell 300 architecture from Nvidia, which is like the biggest, baddest, newest thing as of mid to late 2025. Vera Rubin will

36:37be deployed by the end of this year and models will start training by the end of this year. Yeah. If we are to believe how incredible Mythos Mythos is, I I can't stop thinking about what's going to be possible with Claude code when we have Mythos 1.2, Mythos 1.5, Mythos 2.0, and they do more reinforcement learning on this class of model trained on these GPUs. Like, it's going to be crazy. It's it's going to be crazy. And if they couple that with two, three, four, five, 10 million context windows, um I don't know I don't know what we won't be able to build. I'm I'm very excited for it. That that's That is the only thing on top of my mind except for Pod Sips is

37:23what's going to happen whenever we have more GB300 trained models out in the world. It's It's going to be wild. What do you predict the a strategy to say to stay relevant in that era of AI? Cuz if we can all build at the speed of light and it makes no mistakes, what do you think is the best thing that people should focus on like as far as skills? Like what what should we learn to I think be in business to stay in business? Totally. I think you are the the the quintessential person for this, Bo. Is if everybody can build, nothing matters about building. Building does not matter. If I can pull Pod Sips out of thin air, even if it took me 2 years and I had to learn about vector databases and how to build a podcast episode

38:08processing pipeline, who cares? The thing that matters is how do you utilize this technology to get an audience that pays attention to you, that then wants to download your app. And it's something that I'm not great at, but you are great at. So, that would be my advice is, "Okay, you know that you have Let's just put it this way. You have a sniper rifle that's always going to hit a bull's eye. How do you take that sniper rifle that's always going to hit a bull's eye and use it to help you create content about you always hitting a bull's eye so that you can get people to pay attention to whatever you've built." So, I think content creation is is all that matters and getting an audience and attracting people that are interested in what you've built is all that matters and it's all that these models should be

38:54used for on the cutting edge. Yeah, especially with all the content being AI produced nowadays. I'm pretty sure we've all seen the chart where there's more AI media going out than original. And I think there's going to be that what what goes up must come down, where it's going to now be more valuable things like, "Hey, we're not AI. We're not AI. I'm a real person." You know, I used to have to put that in my videos cuz people think it's AI. Yeah. I did. Remember I texted you and I was like, "Is this HeyGen, Bo?" Every comment has it, so I have to like do something crazy or like fall off my my treadmill or something or make like The three fingers in front of your face. Three fingers in front of your face is a thing. actually broken, so hey, there's your sign. So, hey, HeyGen would have me with a non-broken finger, right? Um but

39:40as the AI gets better, they will be hard to, you know, tell. But I think going back to creating the competitive mode and being like the founder-led marketing, you know, we we see it with other, you know, My First Million, Greg does it, you know, they all it's like people really love seeing the original person, they get to know the person, and they always go back to wow, it's like I actually know you, but I've never met you. And that's going to I think increase in value because there's going to be more demand for it and less supply. Yeah. Like you've seen I have seen in the AI space all of these AI avatars, and they have they're growing like 10,000 views, 30,000 views. They have 100,000 subscribers, but it's all AI avatar. I just feel like there's no connection. Like you're got subscribers, but do they

40:26care about you? I don't know. I feel like connection itself is going to become more valuable. Yeah. I'm even thinking about I'd love to do in-person workshops, like do four a year, go all throughout the country and host these workshops targeted for people that are just like, you know, I I own a restaurant, but I don't know how to implement AI, you know, I'm a real estate agent, I don't know how to use AI. I like that, you know, and I not just from a business perspective, but from helping people hands-on learn how to install OpenClaw, learn how to use CloudCode, cuz I don't see it being done anywhere else. I know that >> Yeah. school system ain't doing it, but, you know, like I think cool would be like that. I love that meeting people and, you know, working together in real life. Totally. In real life events are huge. I I totally agree, Bo, and I

41:12I think you would be fantastic at that. Um, I think >> [laughter] >> I think I think the person that is the best at striking this balance between like AI avatar, it's all AI all the way down, there's no there's, you know, there's no humanity to it is the guy Rowan Cheung that runs The Rundown. Like his whole thing on Instagram is it's only his AI avatar. But all of the production and all of the creativity behind his videos, like you as a human can feel that there's a very clear human touch to it all. So it might be AI, but it's AI directed by a human. And I think that's going to be the thing when it comes to using these models, using the

41:58cutting-edge AI to your advantage for content creation is to take a page out of his book and figure out how to use the models to the best of your ability, but still putting a significant human touch on top of it because otherwise it's just going to be slop. Until it's not, but it'll just be slop for now. Yeah, I think like out out of the box, you know, like if you've ever created a HeyGen avatar, out of the box it's just it's still pretty much like you can tell, but through like tweaking it and tailoring it. I don't know if he has it access a database or or whatnot, but yeah, it's like using it in combination with what you're already doing can be helpful. Yeah, there's another app called Tavis

42:44that uses Gaussian Splatting for their AI avatar. It does a really good job, but um it's still like like I can talk I made one and I can talk to but it's just like there's still like a latency, still kind of like one eye's poking out that way and it's just like But yeah, it's interesting cuz I think from like a coaching perspective, if you could build out an AI avatar and it's available 24/7, if someone just has a question, they can, you know, get all of your knowledge base and access you and get that question answered and like, okay, that's cool, you know, versus like like think about our age of technology today, we still call businesses and are we're on hold waiting to just to ask a question. Like that whole system's being

43:30changed completely. >> Yeah. You know, I think people who use AI to create content and things like that, I think there's a fine line. Like I really do I think there's a fine line cuz everyone's going to be able to do that before you know it. Have you have you played around at all, Bo, for your community and for the people that you're helping? Have you played around or heard of Delphi, d e l p h i? Mhm. No? So Delphi is I've never uh-huh, don't know what it is. >> Yeah. So I have this rule that if someone if the mainly Twitter, Twitter and YouTube, if the Twitter and YouTube AI community talks about something for more than 10 days, then it's worth me looking into. But if it's just like somebody of releases something and it someone doesn't talk about it for more than 10 days, the community doesn't talk

44:15about it for more than 10 days, then it's not me not worth me looking into. Delphi has been around and people have been bringing it up more and more for the last like year or so. Their whole schtick is we will take your content and we will create a like immortal version of yourself that your that your people, your followers can talk to. And I'm actually I was going to build my own rag pipeline ingestion embedding pipeline for a client, but I was like, they've already taken care of this. We should just use Delphi. A consulting client of mine that we're making effectively a digital twin of their brain for their very large coaching mastermind to interact with her. So this Delphi, d e l p h i dot a i is

45:03is undoubtedly worth checking out. >> Yeah, I that's pretty cool. I'm going to I'm going to look into that. Um, I did Tavis, which is just like a human-like avatar to talk to. It still needs work. But yeah, so what what is this? Is it like an avatar? Is it like like an avatar or is it like a chatbot? So it's only a chatbot. There's no there's no visual element to it, but they do have on their I think their more expensive plan, I think it's $300 a month. They have the like yeah, premium pro voice and calling plus chat. So you you load everything that you possibly can into it. I think the premium allows you to put 12 million tokens in or 12 million words. So that could be a

45:49bunch of podcasts, it could be a bunch of blog articles, it could be a lot of stuff. And then they have figured out the rag embedding pipeline to make all of that accessible whenever someone chats with it or has a phone call with it. And they're probably using some low latency phone call system, but anyway, I I'm I'm going to use this for a consulting client that reached out to me and wants to build something like this so that her community members can chat with her Delphi anytime they want inside of Slack cuz she her community is run through Slack. Yeah, see that's that's a really good use case for creating, you know, some type of not like solely for content, but for people that you have clients and you can't meet all their

46:34needs, you can >> Yeah. essentially give it your entire knowledge base and who you are. Yeah, that's that's really cool. I mean that there's a whole business around it, right? Like people I mean I've seen more people pass away and they can digitize their whole life and while that's a little creepy and weird, there's multi-billion dollar startups doing that exact thing. Yeah. So yeah, it's it's pretty interesting where we're headed, where we're going. It's always good to connect with you cuz you just have a different perspective on, you know, what you're building and how you see things. >> Yeah, and I love it. We got to have you on again and just see where your app's at, what you're working on, and go from there. And like I said, people that are watching this on the replay, just check it out. I'll put the link in the description. I'll also put the the developer link, too, if people

47:21wanted to actually hook up their OpenClaw or whatever to it and get going. >> Thanks, Bo. And don't crash the app on the first day. We'll we'll, you know, hopefully that's a good sign though, right? It's a good sign. Yeah, so we'll we'll get you back on the podcast and, you know, if you guys have questions, leave any comments. Do you have any last words or maybe words of encouragement for the builders out there? Um, the the models are insanely good today. Go build something, create a bunch of content around it, and watch your life change. I need to take a page out of that book for myself when it comes to content creation, but you can build anything, so go do it. Yeah, yeah, that's that's the best part. I mean that's the hard part is staying focused, but yeah, I love that. Yeah, beautiful time to be alive. We'll see you guys in

48:07the next episode. I'm releasing a video in like an hour, so have a blessed day. Bye-bye. Thanks, Bo. You bet.