Building Autonomous Companies With AI Agents

00:00 | On this episode of the show, I have with me Ben Cera. 00:03 | He is the founder of Polsia, and we're gonna be talking about, well, company AI run, and how does that actually look in the world? 00:11 | What are these agents doing? 00:13 | Where does the human sit? 00:14 | And obviously, Ben has a fantastic background with being a founder, and we'll talk to him about what the difference is with being an AI native founder and a couple other things in there as well. 00:22 | But Ben, thanks for taking the time. 00:24 | Yeah, thanks for having me. 00:26 | Absolutely. 00:26 | All right, Polsia, what do you guys do over there? 00:29 | So, Polsia is an AI that builds and runs companies autonomously. 00:34 | Uh, you give Polsia your idea, your business idea, and Polsia will go ahead and execute on the coding, the product, fixing bugs, but also on marketing, like, uh, support, uh, running ads, um, you know, tw- tweeting, tweeting on social media, um, et cetera, et cetera, and can also do competitive research. 00:57 | And the uniqueness of Polsia is that, like, uh, it has access to a lot of infrastructure by default, so you don't have to, like, give it API keys or, like, setting set up a, you know, a Vercel or, like, uh, set up a database. 01:09 | It just is very opinionated into having its own setup. 01:12 | And also, every night it w- it does work, uh, in a night shift, and wakes in-- and when you wake up in the morning, you have an email from Polsia that, like, sums up what it has done and what it plans to do next, and what's the state of the business. 01:26 | So you sort of have a, a co-founder that never gives up on your idea and executes on your behalf, uh, on anything that you give it to. 01:34 | You can obviously guide it or l- let it run and rip on your business idea. 01:39 | Uh, it's, it's sort of like a, a, a very simplified way to start a business for non-technical people. 01:47 | Very cool. 01:47 | I, I, and we're gonna get to it 'cause I think, I'm sure most people will be like, "Hey, Amir," you know, "spend more time there." But, uh, before we do, I, I just wanna set some context. 01:55 | So obviously you've been a, a serial founder. 01:58 | I mean, you found some, uh, companies that, uh, are actually well-known brands as well. 02:04 | Obviously, Polsia, we'll talk about that. 02:06 | But in, in those cases, just to kinda get your feel on being this founder and now actually being a founder of a AI native product, w- i-in that sense, as, uh, from a founder sense, does anything change? 02:23 | Does anything-- has anything changed for how you are viewing-- obviously, you're building companies without any people, so that, that might be, uh, that might be what we're talking about. 02:32 | But when you talk about being a founder still, w-what are you seeing having AI as a partner versus y-the other startups you've built? 02:43 | I mean, I think what's very unique with AI is that it's, you know, we're approaching the state where, like, AI are as good as humans at a lot of tasks. 02:53 | Um, but obviously, instead of having to deal with a team member that you have to, like, teach something and, like, have them execute, you can create agents that, like, will follow exactly your int- your, your instructions, will, will be very consistent in, in their execution. 03:10 | Um, and, and you can modify as you, as you will, and also they're much more cost efficient. 03:16 | And so I think a big part of, like, how I built Polsia, and obviously Polsia itself as a product gives the promise of, of automating 80% of the work and, like, really helping the founder, uh, execute a lot of those business functions that usually would require humans. 03:32 | But also, as I run Polsia, I'm always asking myself, like, "How can I automate more and more," right? 03:37 | Uh, obviously there's still stuff I do manually, like, especially in the marketing side of things, uh, where, like, you know, I try to actually tell a very human-centric storytelling. 03:46 | Like, you know, I do videos, video- videography. 03:49 | I do a lot of, like, you know, all the tweeting is myself, all that things, you know. 03:52 | All this stuff is, like, really trying to tell a story because I think that's where AI is actually pretty bad at today. 03:59 | Um, but for everything else, it's like, "How do I throw AI at it," right? 04:03 | And for example, one example is, like, support, right? 04:06 | It's like, you know, the customers have issues. 04:08 | How do you set up a support function that, like, can either do refunds or, like, or, or give out credits, but also, like, fix people's issues, understand them, and have... be reactive? 04:20 | Which, you know, I could hire a bunch of contractors to do so, but, like, they're not gonna solve the core issue. 04:25 | They're just gonna reply probably the same way as an AI. 04:29 | But with AI, I can actually structure it in a way that, like, has a, has a pipeline where, like, it really identifies the actual bugs, uh, is very consistent in the way it refunds or not, and have t- follows a clear policy, is very-- uh, and also can identify the actual issues that customers have and create sort of like a bug pipeline that it can fix directly on the platform, right? 04:51 | So you can literally create, uh, a system that is actually gonna be 10X better than, like, a real human team just because you can infinitely configure it, it can actually fix customers' issues in production. 05:04 | Uh, it can be-- it's much more cost efficient, which means you can be more generous in refunds than, like, if you had, like, to pay humans in terms of a margin perspective. 05:13 | So that's an example where, like, it's a, it's a, it's a fascinating world that turns something very boring to support into something that, like, could be critical in making a self-improving software, right? 05:24 | Because support becomes where you understand what customers want to change on the platform to make it better for them. 05:31 | Um, so I find it pretty interesting. 05:33 | That is interesting. 05:34 | Uh, I guess if you were looking at a traditional startup, uh, obviously you've... 05:40 | y-you-your product is not also your business, but it's also a product for you mentioned non-technical people can use. 05:47 | But I guess if we were looking at just who a founder is and the traditional role they've held in these notion that you have to wear multiple hats, you have to work 90, 100 hours a week, all these different things that we've talked about, obviously a lot of it is constrained with time and one person, uh, being the bottleneck. 06:07 | When you look at a founder and you're like, "I'm going to give you back time," w-what kind of proportion of time are we talking? 06:15 | Are, are-- is it minutes, hours, days? 06:19 | I mean, I mean, uh, first of all, like, you know, I'm-- I don't have an executive team. 06:25 | I don't have, like, full-time employees. 06:27 | I don't have an office with employees in the traditional sense. 06:30 | I do work with people. 06:31 | I work with companies, and so I found proxies. 06:34 | But, like, the fact that I can, can still hang, I mean, it's painful because everything's on my shoulders, but I can still hang because I have all the support systems, and like I-- and U- AI maxing is quite incredible at that stage, right? 06:47 | That I'm able to like, uh, you know, have a company that runs almost ten million run rate, uh, with like, you know, over ten thousand paying customers, and, like, it doesn't completely melt with one person is, is sort of a testament to that. 06:58 | I think when it comes to, like, the founders that I'm serving, which is to me a-- to me, I have this, this concept of, like, everyone is a founder, right? 07:09 | Deep down, everyone's a founder. 07:10 | And also founding a company is not about starting and building our company. 07:14 | Founding a, a company is, like, really acting in the economy, like being an actor. 07:19 | Actually, artists are founders, right? 07:21 | Whenever you create and you put out stuff in the world, you really are a founder. 07:25 | If you open a restaurant, you are a founder, right? 07:28 | A founder is not reserved to, like, a software tech Silicon Valley people, right? 07:32 | Uh, and so I think, I think there's a lot of barriers, though, in starting something because it's pretty intimidating, right? 07:40 | Because as you said, you have to wear multiple hats. 07:42 | You have to work 100 hour weeks because you have to figure out so many things. 07:47 | But it's so gratifying, right? 07:48 | When you're founding, because first of all, you learn so much about, about business, you learn a lot, much about life, about different aspects of the economy, which are pretty interesting on, on, on theirselves. 08:00 | And I think a product-- You know, AI can be used in very, very different ways. 08:05 | I think the way I use AI on-- for my users, uh, which is like turning them into founders and not trying to tell them, "Hey, you're gonna get rich." It's more like, "Hey, you're gonna become a founder." And being a founder is not about getting rich, it's about, it's about suddenly being a creator, creating, right? 08:22 | Being-- having an idea and being able to put it into place, right? 08:26 | Which is the-- that's the hardest part. 08:28 | Like I, I was talking to some of my customers, and they're like, "I had this idea for 2 years, but I never got to it because I have a full-time job, and like this was a side business idea. 08:37 | And then like I was-- I need, I need a CTO, and then I couldn't find them. 08:41 | And then I was-- Then at one point I did an MVP with a friend, but then I don't know how to market it." And like Polsia, it kind of like it does, it does the V1 of all this stuff, right? 08:50 | It like build the product. 08:51 | It like connected Stripe, which I don't even know where to start to set that up. 08:55 | Um, it, it like I had like I sort of made a ads, like it started like getting users and then it started sending emails to people. 09:04 | It's like it started doing the thing. 09:06 | And of course, Polsia is not perfect by any means, and like it's, it's a lot of work to make all those agents work at scale, and that's the big challenge that like I'm facing. 09:15 | And I think when I crack it, how-- as I crack it, this is like the mode of like figuring this stuff out, which is very few people are figuring it out at, at, at the consumer scale. 09:25 | But I think you turn people into contributors and like, and like the first idea, like, okay, after a few months, they're like, "Okay, like my first idea, it's, you know, I really were, was able to try it, but like I actually have 5 other ideas." And so then it's like, then they become founders. 09:38 | Like, you know me, it's my twentieth idea or something Polsia that like I executed, right? 09:44 | 10, 10 to twentieth idea. 09:46 | And it's like, and I've been building for a decade, and it's like the idea, the concept of being able to build ideas much more seamlessly and to enable people that are very creative but like were faced with those barriers, either mental, either actual pragmatic of like not being able to execute what they had in mind. 10:03 | Taking that, those barriers away, I think is gonna be a great contributor to a more fair, a more diverse, a more interesting society, right? 10:14 | Where, where everyone can start understanding what it means to contribute and then, and then will contribute, will not all be successful, but it will create more success stories in a more diverse sort of like internet and like economy. 10:31 | Yeah, it's interesting 'cause, um, this notion of founders has be-- I mean, the founder has been glorified. 10:37 | Like there is a, there's a notion of a founder and people have in their, their minds, and then what you just described as creators, founders, someone that can pursue something on their own, pay their bills by it, and be, be maybe more happy, which is something we-- I mean, I actually have a neighbor, he loves to fly, uh, airplanes. 11:00 | He builds airplanes. 11:01 | I joke with him. 11:02 | He's an engineer. 11:03 | I was like, "Man, if, if, if you turned this into business, I, I, I don't know the first thing about it, you probably would make more money." 'Cause he, he does it all soup to nuts. 11:11 | He bu- I mean, his whole garage is full of just... 11:13 | And he's like, well, as you're talking, I'm like, here's a case of a guy whose hobby probably would be his, you know, what, what pays his bills. 11:21 | He just can't do it 'cause he's, you know, he's got a wife and kids, and he's got responsibilities. 11:24 | So it's very interesting. 11:26 | I, I guess a question for you, and I, I know we-- I said I'm gonna cover some parts, but, but this was interesting. 11:31 | You touched on the founder. 11:34 | Obviously, if you're a solo founder or if you're any founder at any kind of business owner, there is a area of being a founder where it's sometimes hard to find answers 'cause you don't know who to ask, right? 11:48 | And, and I know founders that go to AIs that are out there, and they will prompt them and try to get strategic advice. 11:54 | I, I don't know if they're, they're built for that necessarily, 'cause the advice seems to sometimes run in circular fashions. 12:00 | I've been guilty of asking advice myself, I can tell. 12:03 | When you're thinking about a founder in this case and you're thinking about your product in that co-founder hat, being able to help push through those problems, some s-- an AI to be a partner, a true partner, and understand your business and your goals and, and help you step by... 12:20 | I'm trying to wrap my head around just that opportunity, 'cause I feel like that in of itself is, is, is gin- is gin- is ginormous 12:30 | Totally. 12:31 | I mean, Polsia is like configured to be a co-founder and to want you to win versus being an assistant that just tries to reply to whatever you tell it, right? 12:43 | At least it's configured that way, and I think I'm gonna make more and more efforts to make it more, more so. 12:48 | 'Cause I think there's a big portion when I talk to my customers, they're like, you know, they, they like it for many different ways. 12:53 | I think the main way is that like it's just like everything is set up for you. 12:56 | Uh, the fact-- But also it's like they say, "Well, Polsia like teaches me how to approach problems, gives me ideas of what to do to achieve goals," right? 13:06 | It's because Polsia is configured to really help you win and, and I give it a lot of like also it has a lot of knowledge about how to do marketing better, how to create products better. 13:17 | Um, and there's a lot of improvements to do in, in, to make it better and better and like, you know, understanding how to build a, a long product roadmap, like create, create more marketing strategies that I can execute on. 13:28 | But I think, I think what, you know, at least, you know, I've been building for a decade and I, I pass-- I spent the past year just talking to AI to build, and honestly, like I think most, most, you know, end of the day, to be a good founder, you have to trust your gut and you have to be willing to make mistakes. 13:47 | And I think if you trust your gut, you just essentially have to talk to an AI and like lis- you wanna listen to the things that what the AI said that resonate with you, right? 13:55 | Because end of the day, uh, but I think AI is really good at like giving you the options or giving you dissecting like what are the paths forward. 14:04 | And I think at the end day it's about you being able to say, "You know what? 14:07 | Let's go with option A." 14:09 | And then when you go to option A, it's like since AI does the work right away and gets you results pretty soon, the feedback loop of making the mistake or finding something that works is much faster, right? 14:20 | And so it sort of educates the founder in their own perspective of like it's not like AI is gonna one shot just figure out everything, but it's gonna give you the options, give you the techniques, the paths, and then you able to figure out what you think is the best path and, and get the learnings fast. 14:39 | Uh, and, and I think that's very powerful 'cause again, like, you know, I'm configuring more and more Polsia to just turn more people into trying to be founders. 14:48 | Like I'm, I'm launching a freemium version where like you'll be able to experience the product for free. 14:52 | It used to be $50 or nothing. 14:54 | Now there's gonna be a free version that like, you know, of course you can pay more to get it to do more and go faster. 15:00 | But the idea is like, hey, I want to most, more, most people to experience having a co-founder. 15:06 | Yes, it's not maybe a human would be better, but like, okay, cool, but like there's no human. 15:10 | Also, humans are expensive, right? 15:12 | Or they either cost you fifty percent of your equity and headaches if it doesn't work out, or you have to pay them, and then it's like, okay, like then you have to find the capital to find a full-time employee or like, uh, you know, to pay someone. 15:22 | And so I think it's just a good proxy to get started, and it-- is it the end all be all? 15:27 | Is Puls- you know, is the AI, is Polsia the end all be all where like you'll be able to scale a billion-dollar business eventually? 15:34 | I hope so, but like that's a TBD, right? 15:37 | But at least if you can get to like at least, uh, turn people that are not founders into f- uh, to be able to say proudly, "Yeah, I'm a founder. 15:45 | Like I'm, you know, I'm trying out an idea on Polsia, and like, yeah, I got the website ready. 15:49 | I, I got a first customer. 15:51 | Like I'm figuring out if I can scale," you know? 15:53 | Well, you, you, you, you went pretty far, and like probably once you had the experience once and you probably spend like $100 on Polsia or $50 or whatever it is, it's like, well, now you have the next decade to decide the other ideas you wanna do. 16:09 | Or like maybe, and maybe AI gets better, and maybe you then want to start a business with someone else because now you have all this knowledge. 16:16 | So I think it just creates more contributors in the, in the, in society, I think, right? 16:21 | And you kinda want more contributors. 16:24 | And I think everyone deep down, and that's the American dream, right? 16:27 | The American dream is like to start from nothing and like build stuff and, and, and figure out a, an, a sustainable income. 16:34 | Like be, you know. 16:35 | That's-- I think that's what the American dream is, right? 16:37 | It's like s- this idea of success by creating and working and grinding. 16:43 | And I think that like it's a dream, it's a collective dream for a reason, because everyone looks up to it, I think. 16:51 | And so I think that like once people, instead of feeling like they can't, they are complaining that they can't access it because of inequalities and stuff, which are equality... 17:00 | Inequalities are cr- are true. 17:02 | But if there's a tool that like says, "Okay, you c-- yes, there's inequalities, yes, but at the same time, why don't you try to build something?" 17:09 | And I think it gets addicting, and I think, and it doesn't have to be, "Oh, start an AI SaaS." It could be like whatever, like, "Let me help you figure out your plane situation. 17:18 | Like I'm not gonna make the planes, 17:20 | but like I can help you set up a little site where like you can try to find people to buy your thing, and I'm gonna set up the payments rails, and then I'm gonna try to give you some hints into how to strategize getting customers or..." 17:32 | You still have to do the offline work of like, you know, building the plane and like figuring out what you sell and then-- But maybe you can, you know, you-- that guy could at least try to like be like, "Hey, let me try on the side-" 17:45 | Mm-hmm. 17:45 | "... to try to turn into a business." And I think that's powerful. 17:48 | I mean, reducing the friction and helping people fail faster. 17:52 | I mean, a-as you were speaking, I was thinking about those two dynamics and why a lot of people say, "Oh, I have these ideas," and years go by and they see somebody else do it and they, "Oh, I had that idea. 18:04 | I mean, like I didn't..." It's, i-ideas are ideas. 18:07 | It's the execution. 18:07 | It's interesting 'cause as you're saying fail faster, lower the cost of failure. 18:12 | I mean, these are all the things that, you know, for, for decades people have mentioned as, as why they don't start businesses. 18:20 | I, I guess f-for, for Polsia and how you envision it, I mean, obviously you mentioned where it touches the different areas it could touch. 18:28 | A-a-and for somebody that, that doesn't know The- these different agents, how do they coordinate? 18:34 | So obviously, if they're a co-founder, they're providing, there's gotta be some coordination, some understanding. 18:38 | Talk to us a little bit about how, you know, support and then marketing kind of work hand-in-hand, because those are probably 2 big areas people have issues with when they start businesses. 18:49 | So the way to think about Polsia is, um, there is a bunch of different agents, uh, that all have different tools that they are-- they have access to, and they have sort of skills, you know, like, you know, prompts or, like, knowledge about how to execute what they are supposed to execute. 19:09 | And increasingly, they, they are in a workflow where, like, they are sort of a recipe of, like, how to execute different tasks, right? 19:18 | And, and the way they get called is either by the human themselves who tells Polsia, "Hey, like, let's think through what we should do. 19:27 | Like, I think I wanna do this," and Polsia's like, "Cool, awesome. 19:29 | Well, let's create a task to call this other agent to run that task." But so maybe it's like, "Hey, I, I wanna build a landing page or, like, a site that, like, I can collect, you know, emails from prospective customers," and it's like, okay, well, Polsia's like, "Cool. 19:43 | Let's call the engineering agent to, like, build this." 19:46 | Uh, or, "Hey, I wanna start getting some customers. 19:48 | What do you think? 19:49 | What could we do?" "Well, we could do ads, or we could do cold outreach." It's like, "Okay, let's do cold outreach because that's cheaper." 19:54 | "Cool. 19:54 | Let me call the specialized agent cold outreach that has access to email and has access to, like, a professional database of emails, that has the-- some guardrails about what to do, maybe asks you to set up a custom domain to do the cold outreach," like, all the steps, right? 20:08 | Um, and so there's all these different agents, and, like, either the hu- the human is like, "Hey, I want you to do this," and it does it, or the AI itself every night as it decides what to work on, decides itself, "Okay, based on the state of the business of where we are, is the product working? 20:25 | Like, do we already have customers or not? 20:26 | Like, what have we tried already?" 20:28 | And decides to trigger one task or another. 20:31 | Like, "Okay, well, maybe let's fix a bug," or like, "Let's-- Oh, there's a customer that's contacted us, let's reply to the email," or, right? 20:38 | And then depending on your plan, it can work more and more. 20:40 | It can do more and more tasks during that night shift. 20:43 | So that-that's where the AI sort of decides what to do next. 20:45 | And there's a third mode, which is, like, I call God mode, which is, like, essentially, um, Polsia autonomously decides for an hour to f-5 hours to 10 hours to 12 and 24 hours, depending on how much you, you configure it. 20:59 | It can run continuously. 21:01 | So, like, itself, it just decides what to do next on its own based on where it's at. 21:06 | So similar to the night shift autonomy. 21:08 | So essentially, it's like different decision-makers of what to trigger. 21:13 | So it's like, what should I do next? 21:14 | So either it's a human or it's the AI itself deciding. 21:19 | Uh, I-I guess when you-- That's fa- that's fantastic. 21:21 | The, the accountability of the AI, I, I mean, a lot is made about people worrying about the AI being right, being wrong. 21:29 | We as humans are wrong a, a lot, uh, uh, and often. 21:32 | Uh, when you think about a, a solution such as yours and you're like, "Listen, let's say it's right X percentage of the time," I, I-- my philosophy is it's better to action on something that's 90% right than go, "Oh, I'm not sure," and that reduces my capacity, my ability to, to get to this point A that you're trying to get people to. 21:55 | But talk to us a little bit about when it comes to human error and AI error, and obviously, you have people using the product. 22:03 | I mean, what is their experience with them being okay with the AI making a mistake here and there? 22:10 | Totally. 22:12 | I mean, I think that, like, the way I, I-- the way I personally look at, like, Polsia, it's sort of like if you hired a team, right? 22:19 | Or you hired an employee, and you were like, "Hey, like, go, go do this." 22:23 | And then the next day you check in and you're like, "Wait, wait, wait. 22:26 | Why did you do this? 22:27 | Like, why did you make-- Why did you introduce this bug?" Or like, "Why did you send that email? 22:32 | I didn't want you to send that email." And then you-- ideally, you can self-correct it or tell it, "Hey, next time don't do this," and, like, it remembers, and the next time it doesn't do it. 22:39 | And so I think that, like, that's the first thing. 22:42 | It's like s-similar here, right? 22:43 | It's like, and the AI has guardrails itself. 22:45 | Like, I did a lot of compliance and security things that it doesn't do anything wrong, doesn't do anything shady. 22:51 | And also there's not-- it's not like today, at least it doesn't have the customer's credit card to spend. 22:55 | It's like, "Hey, I spent $1,000. 22:57 | What did you just do?" Right? 22:58 | It's like, it's very contained, and, like, it's like the user knows how much they spent, and, like, they put money first, and then it does, right? 23:05 | Um, but the way, the way I think about, about it is that the product is configured in a way that, like, by default, it does one task every night, but, like, it's gonna do the tasks. 23:19 | It's gonna look at, like, what the user wanted to do, and then if it thinks what the user wants to do on the task that, like, were created, it will try to pick a task that the user wanted to do. 23:28 | So that, like, the AI sort of, like, goes along with what directions of the user s- the user has if the user is do- giving instructions. 23:38 | If the user is not giving instructions, then it will decide on its own. 23:41 | Then that's the beauty of Polsia. 23:42 | It's autonomous. 23:43 | You can sit-- You can say, "You know what? 23:44 | This week I'm busy. 23:45 | I'm on vacation." 23:47 | And Polsia's like, "Well, I don't know. 23:47 | Like, no one's talking to me, so, like, I guess I'm gonna move forward," right? 23:51 | But if you had said, like, "I don't want you to do this," it, it, you know, it remembers, and then, like, it's not gonna do that specific thing that you, you said it doesn't-- you don't want it to do, right? 24:01 | But I think there's also, like, a product like, you know, the God mode where, like, you, you, you press it, and you just watch it work. 24:07 | And I talk to a lot of customers, and they're like, "Yeah, sometimes it's just like, 'Hey, like, let's, let's see what happens.'" Right? 24:12 | And so they decide, "I'm gonna let you decide for the next 3 hours. 24:16 | Just, just do it," right? 24:18 | And you can steer it. 24:18 | You can give it instructions first and be like, "Hey, I want you to focus on this." And it's like, "Okay, go." 24:24 | And it just does, right? 24:25 | And I think that becomes a user deciding if, like, they wanna press that button or not. 24:30 | Like, they're not forced to do so, right? 24:32 | Uh, and so- I think by default, Polsia is trying to be autonomous first because that's the whole point. 24:39 | Like, if by default it's like Claude Code, like, waiting for your instructions, right? 24:44 | Then the user never is like, "Okay, this-- What is this product? 24:47 | It's the same as every other AI," right? 24:49 | It's like, "I need to do the work." 24:51 | So Polsia is trying to, like, shock people a little bit like, "Hey, l- I'm just gonna do a bunch of stuff for you." 24:56 | Most users are, like, amazed, you know. 24:58 | Zero one point one percent are like, "Why did you tweet this? 25:01 | Delete this right now. 25:02 | This isn't my..." You know? 25:04 | And it's like, "Okay," and then you can delete it right away if you want to. 25:08 | Uh, but I think, yeah, I think that, like... I think that, like, a product like Polsia is pushing the envelope very, very far in a world where, like, it's almost, the AI is almost not ready yet, but you, but you kinda, you gotta explore. 25:25 | And I would, I would really see-- I see Polsia compared to, like, the ChatGPT, the Claude and the Claude Code of the world, the same way, you know, Google and ChatGPT were in 2022. 25:36 | So it's like Google was like, "We are the source of truth of the world. 25:40 | Like, our mission is to provide the world with the correct information." Like, I don't know what their mission statement is. 25:45 | It's probably something like that. 25:46 | And ChatGPT came out and be like-- And G- uh, G- Google had the technology to do, uh, LLMs, right? 25:53 | They actually invented it, and ChatGPT was like, "We don't care. 25:57 | We know it hallucinates. 25:58 | We'll put a disclaimer at the bottom, but the tool is too-- is, is, is correct 80% of the time," or whatever the number is. 26:05 | "And so the 20% of the time, we just put a disclaimer," and, like, and the user doesn't have to use ChatGPT if they don't want to. 26:10 | It's not like we are forcing people to use it. 26:12 | And if they wanna use it, we put a disclaimer, like we, you know, it hallucinates sometimes and, like, but it's so powerful that we should do it. 26:19 | And it's crazy that, like, Google, uh, ChatGPT was able to take-- You know, ChatGPT Anthropic to $2 trillion of value against Google's, like, let's say 4 trillion. 26:28 | Like they-- That's so much value that didn't go to Google, right? 26:31 | I mean, and that's great for competition and everything. 26:35 | And I, I see Polsia in the similar thing. 26:37 | If you look at ChatGPT Codex, nobody does autonomy because it's really strange. 26:42 | That's the next level of strange. 26:43 | It's like, it's like, are they ready to associate their brand with an, with an AI that just does on behalf of users with n- sometimes one-way doors? 26:52 | Like, you know, I send that email, I can't take it back, right? 26:55 | Well, they might not be comfortable with it yet 'cause they're like, "Well, the AI is not there yet. 27:00 | What if this, what if that?" And like, I'm like, "Well, I'm gonna-- The user knows that it's autonomous. 27:06 | It's, like, literally, like, in the, in the tagline." 27:10 | And some people won't like it, but the people who do, they will o-- they will discover a new world of, like, trusting an employee, an AI employee to do stuff and being able to guide it and g- and get to trust it. 27:24 | And when that works, it's extremely powerful because it empowers people that, like-- It's not like, "Oh, we are, we are replacing jobs." 27:31 | No, no, no, 27:32 | There was no job to be created because the person who was, was, was, was using Polsia, they, they couldn't afford an employee. 27:39 | So like, no, you're just creating a, a one job, which is a founder now is in a position to create. 27:45 | But they-- We're not replacing any human because there was new human to begin with. 27:48 | 'Cause there was no company, 'cause there was no team, because it's a new founder that, like, was actually probably an employee or a contractor or, like, someone who, who was doing, you know. 27:58 | Does that make sense? 27:59 | I, I, I love that 'cause I think that's the, um, uh, people talk about, uh, AI and jobs and, um, uh, i-i-in this case, you just articulated the amount of additional work that will be put into society. 28:16 | Like, how that work gets consumed and delivered, I think, I think all throughout civilization, there's been more work produced and more work delivered. 28:24 | So I think as technology has facilitated that, in this case, Polsia is actually giving someone that otherwise would never, ever probably take step one to be a creator, put something out in the society that will then spur a-another part of the supply chain. 28:39 | I mean, it's all a supply chain in the end. 28:40 | It's very-- Ev- It's an AI. 28:42 | It's still somebody, um, that has to be behind the scenes doing something. 28:47 | It can't just-- In this case, we put in God-mode and see, see how far it takes us. 28:50 | But I mean, it's, it's really interesting 'cause as, as you've been talking about control and you've been talking about people and their comfort level, I think, I think-- I mean, we could obviously delve into human psychology, but it's, this is a very interesting thing of every, every start of a technology boom of, of any kind has always been with the skepticism. 29:11 | And it's funny 'cause the ones that are the most in position to be rewarded are the ones that embrace it early. 29:20 | And what you just described, uh, I, I have not used Polsia, so I can't, uh, ex-explain how I feel, but I-- if it gave me a competitive advantage and I was 90% correct, that's not bad. 29:33 | That, that's-- We're all looking for a quarter step, uh, of a competitive advantage and, um, it's really fascinating what you've, what you've put out there. 29:42 | Absolutely. 29:43 | Ben, uh, you've gotta go, uh, manage the agents and, uh, and, and keep launching new, uh, features in the world. 29:49 | So I, I know, uh, we gotta wrap up, but I appreciate your time. 29:52 | Thanks for coming on. 29:53 | And, um, if, if somebody wants to learn more, I mean, you covered a lot of ground for us, but what is a good way of learning more? 30:01 | Yeah, you can follow me on Twitter, uh, @bencera, um, or on LinkedIn as well, Bencera. 30:07 | I think those are the 2 main channels I communicate. 30:09 | And I have a YouTube channel also, uh, where I document- 30:13 | Okay 30:13 | my, my journey. 30:15 | There you go. 30:15 | I love it. 30:16 | Well, hopefully, people will follow and, and get involved. 30:19 | And, uh, again, thanks for taking the time. 30:21 | Appreciate it. 30:21 | And that's it for this episode. 30:23 | Be back again, different guest, different topic. 30:24 | Until then, 2 things. 30:25 | One, great, great story behind what Ben's doing with Polsia, and providing an opportunity to people that otherwise would not have an opportunity is what technology can do. 30:36 | In this case, AI is actually doing it day by day. 30:39 | I, I normally ask you to share this episode with somebody else who would benefit from listening to it. 30:43 | In this case, why don't you share this episode with somebody that you know has always wanted to start a business, wanted a founder? 30:48 | Share it with them. 30:49 | I love it. 30:49 | I'd love it if we actually had them listen to Ben and realize that there is a potential door they can explore. 30:55 | So that'd be fantastic. 30:56 | Also, like, subscribe, comment. 30:58 | Let me know how the show's going for you. 30:59 | Until next time, thank you and goodbye.