You Are Not Thinking Big Enough About AI

The biggest problem that I'm seeing right now is that people are just not thinking big enough. AI is going to be a net positive long term for us. How do we think bigger? It would take eight hours of reading per day for about thirty-six years to read what happened in one month. Why can't we just build the next Google?

The work itself can be performed by these AI agents, but the ideas, the, the taste, the reasons behind what we're doing, that is still what we have to communicate. I don't know that there's a better tool out there for extracting information out of your own mind than being interviewed by AI. Uh, with AI, you can outsource your work, but you can't outsource your understanding. So you're not just someone who is using AI to build. You have this very unique business that you built in Lash Loop where you're actually helping other people build with AI as well.

And if we're gonna have a conversation about AI, which everybody seems to be doing, I think it's important that we kinda level set on, like, where are we right now? Before we start talking about where we can go in the future and what a founder should be doing, shouldn't, how they should be looking at it, how things-- where you see things going down the road. Like, what is the baseline reality of what a founder should expect in implementing AI into their business? And let's, let's assume, let's take two cases here as you answer this question. One is, say, the AI native founder who maybe is, has an idea and is coming to a platform like Lash Loop to actually build their idea from scratch.

And then let's contrast that against what someone who maybe has a more established business and is now trying to bring AI in. What are they-- Like, what are the realities for them on the street, uh, you know, in terms of what they can expect to get out of these tools? Because it seems like there is so much... I'm gonna use the word propaganda, but not necessarily in a nefarious sense. It just seems like everyone kinda sits on whatever their bias is and then just projects down the mountain.

And I'd love to, as much as we can, just have an honest level set, and then we can push into our biases as we go. But, you know, wh-where are you seeing the world today and what is actually possible with AI? Yeah. Uh, great questions, and it's great that you separated those, the AI native business versus the existing business too because I think there's differences. Um, so real quick backstory is we started in, like, twenty twenty-two working with these ideas that were not yet possible, and then, uh, GPT-4 comes out, and suddenly these ideas were possible.

And so, uh, I feel lucky in a way that we already knew some of the things we wanted to build, and AI made those things possible for us. But, uh, yeah, I've been working on these, like, early AI agent products and things and experimenting and, and trying these ideas. We released the, uh, first autonomous AI coach with Heights AI Coach in our, my Software Heights platform, uh, back in twenty twenty-three. Um, but AI has changed so much since then. And so where we're at now is, is kinda like the future is here, but it's just not evenly distributed.

And what I mean by that is we have reached the point with the models that have come out between the very end of last year and right now that they are just so much more capable than they were a year or so ago. And the, the issue is that there are some people who are, like, massively taking advantage of that and are getting so much value from these models, and there's others who feel like, "Oh, is it... This is the same ChatGPT or, or whatever that I was using, uh, about a year ago." And they, they kinda don't realize the differences or, or what's available to them. Um, and so I would say for the, the AI native business, um, I think th-those kind of, uh, founders and creators are kind of figuring this thing out, uh, already.

Uh, whereas the, the existing businesses, uh, I think they have, uh, some things to do about how they... Like, I don't think this is about, like, how, how do I fire half of my team or something like that. I think it's more about how do I change the way that my team thinks of how they work with AI because e-everybody's being, is now able to become kind of more an, of an operator of controlling these AI agents and directing them in a certain way. And, um, yeah, so I guess so real quick of, like, where, where we're at now, um, but I guess to be more concrete on, like, what the models can do now is, uh, I've been seeing my own usage in, like, writing code. I remember at the end of last year, I was using, like, five hundred million tokens per month in, in these coding agents.

And at that time, people were like, "Wow, that's, that's kind of impressive." And I remember some people were surprised by that. Um, in the beginning of this year, it was like a billion tokens per month. And then I remember hitting, like, the next month was two billion. Now it's, like, over three billion.

And the amount that I'm able to use just keeps going up because the models are so good, but the work that I'm putting in is not necessarily more. And so we're suddenly seeing this, like, massive output that, uh, I'm trying to, like, talk with other developers because I don't even know what the baseline is anymore of, like, what is a, a high amount of, like, code changed in production per month or something like this. So you just described a scenario that I think-- So, so first of all, to level set, the audience knows this, you may not, huge AI optimists. So this is something I've been pushing on my socials and stuff a lot because, um, I think that all these AI doomers out there are doing... Everyday users of AI, particularly business owners, will say in the small to medium-sized space, mid-market space in particular, who may not be tech founders or tech-oriented, it doesn't mean they're, uh, Luddites, just, just, you know, it doesn't come naturally, which I would put myself in that scenario.

I believe in technology. I've been around it my entire career, but I was never a coder. I took one C+ class in college and was like, "Nope, this is not for me." So but I appreciate it. So Now, if I wasn't kind of as open-minded to this stuff as maybe I just my natural proclivity, I may buy into this AI, you know, is gonna wreck all jobs, it's gonna remove all satisfaction from work, and people are just gonna be, you know, taking some universal basic income and having no purpose in li- And I'm like, none of that is gonna happen.

None of that is gonna happen. That is all this crazy, almost, like, demonic scenario of, of what AI could be. And I, and I guess it, it's ... There is a percentage chance that it could happen, but it's never happened before, yet the same language that's being used towards AI right now by the doomers was used when the printing press was invented, when the car was invented, when the internet was invented. You know what I mean?

Like, we've been told this over and over and over again. So okay, I'd like to believe that history rhymes and sometimes repeats, and in that case, you know, things will be different, right? The world was different after the car before ... than before the car, different for after the printing press than before. But we're still here, we're flourishing, and, and I honestly believe that. I think what you said, though, that was... that's really interesting, and I, I wanna frame it, and then you, you take this where you will.

You said, "Right now, no one really understands what the baseline consumption versus output of tokens is, what you should be getting as an ROI or as output." Okay. And that says to me that we are living in this kinda wonderful FA/FO moment, where the answer is most likely go out and do it, play around, or, or, you know, make a more serious push, depending on where you are in the, in the curve of, of, of AI adoption understanding. But you gotta be out there in the game, you know, pushing code out and trying to build stuff, even if you never use it in your business, just to understand what it does. But you have to be playing around with this stuff to have a feel for it, and that there isn't really a right or wrong answer today.

Is that a proper way of framing this, do you think? Yeah, I completely agree. I think you have to try out things with these models, and I think the biggest problem that I'm seeing right now is that people are just not thinking big enough. And, like, I realize it's this thing for myself. I have to constantly challenge myself of like, "Well, how could I just think bigger on this and, and do something that before it would've been like, 'Well, this is, like, a year-long effort,' or, 'This is, like, a year-long effort with a team.'"

And now it's like, okay, well, let me try this over the weekend quick with the AI. And so even if you tried something, like, maybe six months ago and AI couldn't do it or AI messed it up, um, yeah, why not try that again now and see, like, okay, if the AI does do it, okay, well, can you, can you think bigger than that? What's something that is more impressive? Can it also do that? And I think people are getting stuck in, like, building this little thing, but not thinking forward to, like, either where does it go from there or what could you actually accomplish from there?

Because I completely agree with you. I don't think, uh, we're, we're all gonna, like, uh, lose our jobs and have nothing to do. I think there's a lot to do, and I think we're underestimating the things that we could be doing now if you have this, this, like, resource of AI that an individual can direct it in so many ways. Yeah. I just saw a article, the CEO of Cognizant, one of the largest, uh, management consulting and tech consulting firms in the world, um, he just came out and said they are actively recruiting 20,000 undergrad graduates, um, because th- they've... what they're doing with AI has created so much additional work, and, like, whether it's, uh, orchestration or human-in-the-loop touchpoints or, um, output validation or all these different things that need to be done by humans, that they're out there recruiting 20,000 new employees.

That's white-collar work, right? And you also think about all the contractors that need to be done to build the infrastructure to build... You know, I mean, what no one's talking about right now that I think is really interesting is, you know, you're consuming three billion tokens or using three billion tokens a month, and from what I heard, it's probably only gonna go up, right? Well, we need more, like, infrastructure in terms of hard wires and, and, and electricity, and that's all gonna need to be done by contractors, and that's not, like, a two-year project. That's, like, a 50-year project.

So, you know, I think about this and I'm like, okay, if we can all agree, and I know many people won't. I, I will get hate on YouTube and in the clips that we pull from this for being an AI optimist. I always do. But my point is if, if, if for the purpose of this conversation, if you guys are listening at home, if you can just, whether you believe it or not, buy for the remainder of the conversation that, that AI is an, is going to be a net positive long term for us, how do we think bigger? 'Cause I love that you said that, and I've actually found in my own work questioning some of my own, um, like, like assumptions in, in what you just said, right? Like, I, like, it, it doesn't take a week or a month or a year to build something.

It can take a couple hours on a weekend to have a, even a functioning prototype of maybe... Like, I built this little connector between my website and this other tool that I wanted my website to, to use. One, it would've never been able to do it before unless there was, like, a WordPress plugin or something, and, you know, I had completely torn my website down and rebuilt it from scratch so that it could be, like, a, a AI native website. And then I built that connector in 45 minutes using, uh, Opus 4.8, right? Like, I just went in, I said, "Here's what I wanna do.

Here's the other system. I want my website once a week to ping this system, pull these results, analyze it, deliver it back to me," right? Like, and it just, broom, builds it, tested it, prototype out the door. It doesn't mean there aren't still iterations to be done, but that was, like, two hours on a Saturday morning where that connection wouldn't even have been possible, or I would've had to use multiple systems or, you know what I mean, all these other options and- That's this tiny little microscopic idea. So if I'm sitting here and I'm looking at my business and I'm going, you know, I'm starting to maybe catalog where some of our friction points are or some of where our, like, hard passes are, where one system doesn't talk to another and a human has to literally pass that information, ho-how do I start thinking bigger about what AI can do for my business?

And let's take the scenario of a preexisting business, not a AI native build from scratch. Yeah, that's a great question. Um, so yeah, I wanna give, uh, also, like, a perspective on, uh... because I completely agree with you. We're gonna just keep using more of all this. I think people are really still underestimating the demand that there will be for the AI usage as the models continue to get better.

Because, like, tho-those who are business owners, those who are, like, at the forefront of trying to build things with these tools are now able to suddenly use, like, way more. But, uh, yeah, I think it's just gonna keep going up. And, like, to give you perspective, uh, I remember when I hit, like, the two billion tokens per month, I tried calculating, like, well, what does that actually mean? And it would mean that if you want to read every single token going in and out of the model, it would take, uh, eight hours of reading per day for about thirty-six years to read- ... what happened in one month. And so it starts to get crazy of, like, where this is going from we send a message to ChatGPT, it sends a response, to now you have these agents that are able to run for, uh, some period of time and actually accomplish work for you.

And so what this, what this looks like, kinda like the roadmap for what I think businesses should be doing and where they should be thinking is, uh... I'll give you one example of, like, where I kinda challenged myself to think bigger recently. Um, with my Business Heights platform, we help, uh, creators and entrepreneurs who are building these online knowledge businesses, a community membership, a course, a coaching offer. And, uh, we've been working internally on this, uh, like, survey that we're trying to put together to try to understand the, the creator economy as it is right now in twenty twenty-six. And so we're pulling data from, like, our platform internally.

Uh, we're making a survey to ask people about and things like this. But I was thinking to myself recently, like, I would love to know about, like, other platforms and competitors and stuff. Even, like, not just to know about competitors, but just to have a broader picture of, like, where things really at as a whole and not be biased by, like, just the kind of creators on my own platform. And so I thought to myself, "Well, why can't I just be a next, uh... Why can't we just build the next Google?

Why can't we just build a Google where we have our own web crawler, web search that's going to build a database, a database of every creator out there and learn all about them, learn what they're doing, and then we can be able to, like, pull data from that and ur-understand, like, okay, the creators who have been around longer, do they have, uh, like, they have this many web pages on their site versus somebody else? And, like, where can we pull interesting information from that?" And before it would've been like, "Okay, well, this is a really, like, complex project." And now it's something that, like, the, the MVP is built already from, like, a couple prompts. Um, and so, like, things like that that you would just never consider, like, even being able to do for your business are now this, like, it's just if you have the idea, like, might as well try it.

And I think that the way that you begin to think these ways, uh, also is that you have to be able to learn to, uh, communicate, like, your intent in, like, the clearest and fastest way possible to get these agents involved in things. Um, but stop thinking o-of it, like, task by task of each little thing. And it's more about now, like, a, a broader, uh, bigger plan. And so, like, the, the kind of prompt that I gave, like, a, an AI agent, uh, for building that kind of search engine was not, like, a couple sentences. Um, it was like a twenty-ish page or so prompt of text of everything that it had to build.

And then I let it do it and just walk away and, and see what happens. And I also didn't have to write the twenty pages, right? So I was communicating with AI, kinda having it interview me to understand what do we actually need to accomplish here, then it wrote the, the twenty pages of its own implementation, and I said, "That looks good. Let's go for it." But, um, I think, uh, like, the, the founders out there need to be thinking for themselves and for how they have their teamwork in the future is, uh, like, designing these, these processes that you can delegate.

And, uh, I'm very happy to see that, like, the last couple days on X, people are talking about loops and that the future of working with these agents is you're designing loops that are going to be running for you in your business. Um, and that's great for me because, uh, our coding agent is called Latch Loop, so ho-hopefully that sticks around. Um, but, uh, the, the, the phrase sticks around, and, uh, people can hook onto it there. But, um, yeah, I think figuring out where you can design the, these broader goals that you want to, like, distribute the attention to, um, so you can have AI working on these kinda bigger picture things that you may have not even considered before. Can you just explain the idea of a loop? 'Cause I saw that as well on X, but I'm sure most of the audience is unfamiliar with what that term means and its, uh, implimica- implications to building.

Yeah, 'cause I remember I was talking with, uh... I went to OpenAI Dev Day last year, and I went to this separate event of, like, uh, devs talking at this, uh, like, other, other people building these AI agents. And I described the, the name Latch Loop of our coding agent to them, and they didn't understand what it was either. Um, but the idea to me is that when in programming, if you have a loop, it's saying like, "Okay, while this thing is true, like, continue and repeat." And so what developers found out is you have a, a tool like ChatGPT, and you can, uh, send it a message, it sends a response.

But if you want it to keep working, you have to have a way for it to continue in a loop and, and work on something. And so these kinda coding agent tools that we see, what they're doing is we're, we're giving them a goal, and then we allow the agent to continue working. So after it edits a piece of code, the system shows that back to it, and then it decides, "Okay, now this is the next thing I'm gonna do. This is the next thing." And some of these tools will even do things where, like, if there's a to-do list, like, the agent has to... is forced by the, the programming To repeat until the to-do list is finished.

Um, so that's the idea of a loop, and, uh, uh, some of the ways that you can do these things, like inside ChatGPT directly or inside these agent tools, is a lot of them have like an automation section. This is the, the easiest way to set something up as a, as a non-programmer. If you can think of a task that you would have repeated, you can have a, a small loop that it r- is repeating daily, weekly, hourly, um, for that. So something could be, um, like find one small bug in my software or something and try to fix it, or find one, uh, file that's getting too long in my software and try to optimize it or make it shorter. And like these are little things that maybe you would wanna spend some time on, um, but like now that AI can do it, you can just have that kind of running on a repeat process, and it's just constantly approving.

It doesn't need your, uh, direct input necessarily. Yeah. And maybe, uh, so I, I use, um ... I set a couple of very simple ones up where to handle email 'cause I don't ... I've tested almost all of the, like, AI email tools, and I've just never, I've never really been happy with them.

I just don't... It ultimately comes down to I don't need all that, and I like working inside of Google's kinda native email system. I have it set up already the way I like and all that kind of stuff. However, there's certain recurring emails that I get that I just don't want to clutter up my inbox, and I know I- you can create certain tasks inside of, uh, Google natively, but, you know, it ends up being you have to have 400 of them because it, it tends to be very like specific one-to-one kinda stuff. And I have, uh, like just for the audience mostly, not this won't be revolutionary for you, but like receipts for my business.

So anytime a receipt comes in, it's scanning my inbox twice a day, once in the morning and once in the evening. It's finding those receipts, tagging them, moving them to a folder, and then forwarding them to my accounting software. Boom. So now the receipts that I get, however many of those come in a week, day, or, or month, et cetera, I never even have to look at them, and if I see one, I know it's ultimately gonna be taken care of, and I can a- just scan past it, and that way I don't have to set up individual rules for every single vendor that sends me a receipt on a weekly or monthly basis. Now the AI is finding it and then et cetera.

So that would be, uh, an example of some, like an automation inside one of these AI tools that you could set up that's fairly basic, but ultimately does create an increase in productivity. Now, what I hear you saying is this actually is something that's very powerful inside a coding agent. So if I'm trying to actually build, uh, let's say I'm trying to build a, a connection between two systems that there isn't necessarily a tool for, or maybe the tool is kinda priced in an analog or digital era style, and I don't wanna pay the $150 a month for it, I could potentially, you know, I could potentially build that connection myself. Uh, you know, you c- you would... What these loops allow you to do, and then push back on where I'm wrong here, I'm just trying to, I'm trying to still make a new case.

Like that loop allows you to, as you described, have the AI... So what I would do is I would have the AI interview me. I might pull up Claude or ChatGPT or whatever my favorite is. I would tell them what I'm trying to do and maybe say, "Hey, interview me to create a plan that I could deliver to a coding agent." Right?

Now that AI is gonna interview me. I'm gonna take that output. I'm gonna deliver it to, say, Latch Loop and a tool like Latch Loop. And now I can give that to Latch Loop and say, "Go," and I don't have to be sitting there now, you know, if a l- if there, if this loop technology is involved, I don't have to be sitting there hitting okay, okay, 'cause I know like the early stages of them, like literally you had to sit there and hit, you know, okay to move on, okay to move on, like even, you know, over and over and over again, and that almost defeats the purpose of the power of these tools. Is that kinda what you're describing?

Uh, completely, yeah. Um, yeah, so w- with the combination of like the agent harnesses, a tool like Latch Loop, Quad Code, Codex, and the model's getting better, now we're at the point that the model can continue towards this goal without having to, you say, "Continue, continue," or, "Okay, okay." Um, and, and so yeah, so it can progress, uh, more deeply on, uh, bigger things. I will say, though, that, uh, I don't wanna go too far in, uh, this direction without addressing that if we think to the future of where all this is going, if you say, "Okay, well, Brian, uh, if we all have these, these magical, uh, AI agents building everything for us, imagine they, they continue to get better and the, our business is being built essentially by these AIs that we're directing, what becomes the difference between my business and your business if we all have the same agents that are running?" And what I would suggest is that business is just how you do things, and if you look at like Apple versus Windows, and like remember the, the Mac versus Windows or Mac versus PC commercials.

And, uh, Apple has always said, "Well, like we have this very specific process of this is the way that we design a product," or, "This is the way that we design software." And so in your business, I think it's very important to identify that for yourself and realize that that's what is, is unique and that's going into all this. So we're not trying to have the AI just generate slop for us. We wanna make sure that we're getting these unique ideas and everything into what we're trying to, uh, like articulate and create. Um, but yeah, like that's, that's the most important thing.

So like the work itself can be performed by these AI agents, but the, the ideas, the, the taste, the reasons behind what we're doing, that is still what we have to communicate. I love that you just used the word taste. I use that all the time. Like when I'm talking to people, I'll say jud- it's judgment and taste. That's gonna be the defining characteristics.

It's like y- yes, you might be building a, a new CRM product for plumbing contractors or something, okay? And there are other people there, but it's w- what is that out-- what is that unique output? What is that unique spin? Just like it was before. It's like I feel like somehow, especially when new technology comes, and we saw this again with the internet, we saw it with APIs, it still comes down to what is the unique idea, whether humans are coding it or Opus four point five or Codex or, you know, whoever's coding it, whatever agent you're using It still comes down to what is that explicit and unique output and your judgment as to why that's important, that look, that feel, maybe it's thinner or slimmer, um, you know, more modern design, or maybe it's, you know, just massive amounts of data that, that, you know, weren't possible before.

Whatever your-- It's, it's that taste and judgment that, as has been the case for the history of humans creating things, that is still gonna define these products even if agents are coding it. I mean, that's, that's what I hear you saying. Is that correct? Yeah, yeah. Um, yeah.

I think what, what we're all doing and where this is going, uh, whether you're building software or something else, is that we're all kind of communicating intent to direct attention. And so before AI, that attention was, like, directing human attention. Like, where are our employees gonna work on something? What is important for us for them to focus on? Now it's on these AI agents and explaining to the agents what are the things that we want them to kinda, kinda essentially spend this attention on.

I wanna come back one more time to this idea of, of not thinking big enough. So for you, when you sit down and you start to vision, you know, kinda map out, we'll say, a new, a new product completely or a new function, a new feature, like, how do you make sure that you are thinking big enough, you know, using your words, you thinking big enough about that thing, that you're, you're pushing the envelope as far as possible with these tools so that you're not just another commoditized, you know, app builder or whatever, right? Like, you have a unique feel. Like, how do you ideate through a pr-- Do you have a process for ideating to make sure you're capturing the full extent of what's possible for this idea that you may have? Yeah.

I, I think it comes back to what we were talking about of, like, just playing with the models and finding out. Um, I, I think, uh, I don't remember if this is the exact quote. Um, I think it was from, uh, Yassine on, uh, on X. And I remember some investors and other people started, uh, quoting it and everything what he said is that, uh, with AI, you can outsource your work, but you can't outsource your understanding. And so it's your job as a human, in order to be able to communicate the things that you have ideas about and the things of where you care about, you have to be able to understand.

And so the, the good thing is you can use AI to help you understand those things faster, but in part, that's from trying things. And so thinking about, okay, well, what if we did this? And it, it's not so much a co- a, a thing of cost anymore of like, okay, well, I can't go and, and spend tens of thousands, hundreds of thousands or whatever dollars and hiring a team to build this thing that I may end up throw out, but now you can just ask AI to do it. And there's still a cost of the tokens, but it's, it's tens or hundreds of dollars instead of hundreds of thousands. And so, um, yeah, it's just like, okay, well, it'd be cool if we could do this, and just try it, see what you get.

You might get something that, okay, actually, this is not there. Why is it not there? Is it because of some technical thing I, I don't understand? Is it something else? And, and whether you're a dev- developer or not, I think you can begin to, to kinda work through these things, um, by, like, taking that process with it.

Yeah. I actually have built three different applications that I have since just blown up or completely deleted. But the process of going through cr-- Like, one of them, um, I really love the name that I came up with, and I got the URL, so I was, like, super excited, but it was this idea of I call... I, I wanted to create, uh, a finance tool for, like, solo entrepreneurs, 'cause I know for myself, I have my personal bank accounts, my personal credit card, and then I have my business bank accounts and my, and my business card. But, like, essentially, you know, they operate in a very similar and very close ecosystem since I'm the only employee in the company as a solopreneur, and, you know, and any contractors I paid, you know, ten nine nine or whatever, but, like, you know, I'm not paying payroll to anyone else except myself.

And then that money is essentially money that, you know, I can use in my personal life. And I was like, it-- there's no real good tool out there for mixing those two sets of finances in a single view, but, but being able to keep them separate in terms of understanding what money is in the business accounts and what money is in the personal accounts. Okay, that was the idea. I called it Black Ink, and I was like, "All right, I'm gonna build this thing for myself, and if it works, hey, maybe there's something here." And I went down the path, and I built this thing out, and it cost me maybe three or four hundred dollars in tokens over the course of a few weeks, you know, putting it together.

It wasn't my primary focus, so, you know, I was taking, taking my time. And I got to the end, and I was like, "This is cool, but there's some pieces here that are pretty complicated, and ultimately, this isn't really a business I wanna be in." And then, um, uh, Perplexity Computer came out with their finance tool, and I was like, "Okay, that's twenty dollars a month." Um, and u- ultimately, I've moved to ChatGPT's new finance tool, which I think is absolutely fantastic, to be honest with you. But I was like, there's better things out here for twenty bucks a month, and I think they're gonna eat this process anyways.

And but it was the process of building it helped me understand what does it actually mean in terms of integrating a pl- Plaid into a business like this? What, you know, what, what kind of, uh, um, security structure do I have in place for them to even give me access to their API, et cetera? What do-- How do I have to map this out? I made a bunch of mistakes because I didn't go deep enough on the, like, what I wanted from the business side in terms of telling the AI, so it kinda came out wonky. Okay, there's a lesson learned.

I didn't map it out or plan it properly. And ultimately, like I said, it was, like, maybe three or four hundred bucks tops, and I ultimately blew it up and decided I didn't wanna do anything with that. But to your point, even though nothing came out of that from, like, a financial or usage standpoint in the long term, I now have a much clearer and richer understanding of what it takes to develop a project from the beginning and what some of these, um, more complicated or more secure connections are gonna cost, what it's gonna take to build to them, what, what they're even gonna allow, what you need to do in ter- and prove to them in order to, for them to even connect to your system, et cetera. And, and that's how you develop this understanding, and it's why I come back to this idea of, like, this is the FAFO moment, like, probably of our generation is right now. And it seems like the people like yourself, like to include myself in there even though I'm far less technical than you, like, even if you don't end up being a hardcore builder of technology, I think taking on some small projects and trying to build some of these things, even if they don't end up working, is going to play- pay massive dividends into the future.

So, you know, I wanna... And, and where, where my question kinda going here is, is this idea, which people have kinda gotten away from this term a little bit, but, like, vibe coding. And I wanna set just a little bit of more context, and then I'll pass it over to you. Um, I was listening to very famous podcast. It was all in podcast, and they had an investor on.

And I don't wanna use his name 'cause I think this guy is brilliant, but he's just hammering on vibe coding, hammering on it. "This is not the future. They're not gonna build relevant applications," on and on and on he's going. Now, if you listen to the full podcast, he then gives away at the end that this is-- he's also a massive investor in Salesforce and HubSpot and all these SaaS tools, right? So he has a vested interest in people not creating technology that competes against them.

And but what, what, what I didn't like about that was if you were considering starting to build your own applications or there was an application you were thinking about building, what he was putting in people's brains is that somehow vibe coding is, is less than, right, or, or is never gonna be equal to the quality of technology that an army of Salesforce developers could create. And maybe, you know, being that you have all this experience, uh, not only with, um, Heights platform, but ultimately with, with Latube as well, and you're seeing people do this in real time, w-where-- you know, what would be your push-- One, would you push back, I guess, on his argument that vibe coding can't produce real functional, commercialized, large scale, uh, applications? And two, you know, I g- we'll, we'll start there. Like, do you agree? Would you push back?

Is it possible to build commercially viable applications for someone like myself, non-non-technical, uh, but I have an idea? Yeah. So I think there's, uh, a, a couple things I think about this. Uh, number one is, uh, I wouldn't suggest that a business go out there and, like, try to replace all their software by vibe coding it and think that's gonna save them some money. Because the reality is that you purchase that software to help you achieve some kind of thing, probably save you some money.

And even if you get, like, version one, uh, done pretty well and you think you're happy with it, most likely the company that's been building that software and has made millions of dollars or has millions of users because of it has fixed so, like, tens of thousands of small problems that people have reported to them and, and figured out or thought of different ways of doing things that you don't wanna have to go, necessarily go through that if you're trying to replace some small little tool that you use occasionally. And so that would be, like, the case against it. However, if you are saying that, like, "I have this, th-this goal that I wanna build something, put it out into the world. I would love to be able to make my own product, but I'm not really technical," um, there are some things that you have to be aware of in terms of, like, security and, and all these things that you, you will have to in-- like, undoubtedly learn certain things in order to be successful if you're not ever planning to have some developer help you. Um, but, uh, you can absolutely do that, and it's, it's such an incredible time to be building something.

But I would kinda go back to what you were saying before about the, the app that you built because I think you touched on something that, uh, is really important and is that, uh, if, uh, if we keep going in the future here and, and all this stuff keeps evolving, uh, where does business go and, and how do we decide what we should even build? What is even worth building? And like you mentioned the thing that you built, now suddenly there's ChatGPT Finance, which is doing it so well. So how do you decide to build a thing that is not just gonna get built by somebody else so easily or, or something like that, right? And I think that where everything is shifting to is towards building for outcomes.

And so, like, building something that delivers the outcome directly instead of just helping to achieve the outcome. And I don't know if you've heard of seeing some people talking about that it's not software as a service anymore, it's service as a software. I think that not only is software moving this direction, but I think even, like, agencies are moving this direction. That, like, so software has to become more like a service. Agencies have to become more like a software in that we're not just delivering something that's gonna, like, people would buy the software because they hope that if they click around the software, it's gonna help them maybe achieve something faster.

Now, there's no reason to learn software anymore. There's no reason to be clicking around software anymore. I can say that as somebody who I'm building the software for a living, right? What people wanna achieve is the outcome. And now with these AI agents, you can build these agents that just help deliver the outcome.

And it doesn't have to be through the software directly and, like, only the software, but it can be the combination of, like, if it is an agency, your team plus the agents that your team is working with in order to deliver that for a client. Okay. So I'm gonna break a scenario down for you, and then you tell me if this is what you're talking about. 'Cause, one, I a hundred percent agree. I think the audience-- I think the idea of service as a software could be a little vexing for some people, maybe just before they wrap their head around it. So, um, I do, I produce, um, a decent amount of content on Instagram in former reels, right?

A lot of it is based on this show. And what I did was I looked at, like, um, Opus Pro, which is a perfectly fine tool. There's a lot of tools that you can use now where you put in some raw footage, and it can spin up a nice, um, a nice out-- a, a nice, uh, clip for you or a nice reel or whatever. But the hard part is a lot of times you're stuck in their templates and that kind of stuff and which is, you know, can be fine, but you end up kinda looking like everyone else, and I wanted a unique flavor. So what I did was I used an agent to talk to, uh, Remotion and a couple other tools, Higgsfield AI, et cetera, and then I gave it the plan for kind of the unique feel that I wanted my clips to have.

And then now all I have to do is drop the raw footage in a folder, tell the agent, you know, launch, and it goes out, reads the clips, pulls them, then goes out to the appropriate tools, comes back. And what I just get is, you know, when, you know, x however much time it takes, you know, sometimes it takes ten minutes, sometimes it takes half hour, depends on, you know, how kinda complicated what I'm asking it to do is, I just get the raw output, right? I didn't have to go in and play around. I didn't have to, you know, add text. I didn't have to do all this stuff that you'd normally have to do in a, in a clip editor.

Uh, I just got the output delivered to me, and then I just upload it, and off you go. Is that kinda what you're talking about as a, an outcome versus using the software kind of thing? Yes, exactly. So, like, imagine, uh, before AI, if you wanted the, the reel done for you, then you have to hire a agency or video editor or something to end up with that final product. Now, uh, we have, yeah, something like Opus is, like, they're trying to deliver the outcome, but yeah, in your case, it wasn't in your voice yet, and so you wanted that specific thing.

Now you have that through the system that you created, and now let's say, like, you could go to same kind of companies that say, like, "Well, we want reels that are gonna be in our voice." Now instead of hiring an agency, they can hire you, and then you're delivering that as the outcome to them. So they don't have to know about the software. They don't have to be paying a team for it, but they're paying directly for the outcome. And where this gets interesting is that's one thing, but now, like, what can you do that was bigger than before?

What can you do that you were just not able to ... Like, how can you deliver to a client or customer at a level that was just impossible before, either because it would just take too much, like, individual time, take too much money or, or something else, but now you can thanks to these AI agents. Yeah. It-- not to, not to pull this kinda clip idea out too far, but, uh, you've probably seen, um, a lot of agencies have kinda spun off a service that's called clip farming, which, for those of you that aren't familiar, is you take maybe this ... We would take the raw output from this conversation that Brian and I are having, you hand it to them, and they don't come back with, like, three clips.

They come back with, like, 300 clips, and then they create all these, um, kind of themed, you know, additional Instagram accounts, and then they, you, you know, they're posting these clips all over. So it looks like your clip is being reposted and shared and moved, not just on your profile, but on, like, 15 profiles. And, uh, I have a buddy who is launching one of these services, and I was talking to him about it, and he's like, "Yeah." He's like, "This would've taken, like, 100, like, um, like, humans to make this happen, like, to, to just the time it would've taken to build out all these things." And now what, you know, he's saying, "Hey, what our agency can do for you is we've used AI to code up systems and workflows, et cetera, that can do this on their own."

And now our agency is saying, "Hey, you just hand us that raw file. We're gonna give you back 300 clips. You don't need to do Opus. You don't need to do this yourself." This is now, you know, they've kind of showing both sides of it, right?

They're able to use AI to build this system to create an outcome, but as an agency, in this case, a marketing agency, the, the, the, their customer isn't getting software, you know, like, like you would if you went to, like, an Opus clip or whatever. You're just getting a folder with 300 clips in it if you want, right? And in, and in their case, they actually publish them for you. So that would be kinda that s- service as a software. You put in the raw file, you wake up the next day, and you have 300 different clips of your last podcast blasted all over the internet.

You didn't have to do anything, right? And they don't ... That's not done solely by software. It's done by, it's done by this marketing agency, but to the, but to the user, to your point, they're just, they just want the outcome. They just want the distribution.

That's all they want. They don't want to have to log into anything. They don't wanna have to go in and edit 15 things. They just wanna produce their, their podcast and then have it distributed. Is that, that kinda wrap, wraps up this outcome-based thing?

Yeah, I think so. I'll, well, I'll give an example of, like, what we're doing right now with Heights Platform. So it, it started as this, this all-in-one course and community software you could build and sell your, your knowledge business products through. And, uh, if you imagine, like, as somebody has to, like, set up a online course or digital product and build a website for and send out emails, um, that was, like, the old days of how this worked. And we have this, uh, system called Heights AI inside it that can help you with some things, but we're, right now we're working on what we're calling Heights AI 3, the kinda next version of this that's gonna be much more agentic and, and proactive in how it can help you.

And so where we're turning this into the service as a software is imagine that you're, you're selling some kind of information product online, and you wake up Monday morning and your AI agent says to you, "Hey, I noticed that you got some more sales on this product, but actually you weren't promoting this product as much as the other ones. So why don't we send out an email newsletter to your audience about this product since it's doing well, and we can send it to this specific segment, and actually, here's an email that I drafted for you." And then you have it all set and ready to go of something that was able to spend attention on the things that you care about in order for helping to, like, grow your business. So you didn't have to click around in the software to figure out, oh, this thing was performing better. Oh, maybe I should do a promotion here because I didn't recently.

Oh, maybe I should do this. And the agent was working on that for you, and so you're just making the decisions to kinda direct it where it should go. Yeah, I love that. You know, and, and I think sometimes, say, traditional service businesses like, uh, my home industry is the insurance industry. Much of my, uh, professional experience is coming up through the property casualty insurance industry.

And, you know, I could see a scenario where, you know, one of the big issues is retention, right? So, so how you make your money in property casualty insurance isn't in selling a new policy. Right? That's oftentimes when you sell a new policy in that space, much to the misunderstanding of the general population, you lose money the first year. So like if I were to sell you home and auto insurance, I would most likely lose money by selling it to you the first year.

Traditionally, you do not make money in that space until somewhere between two and a half to three years from the point that I initially sell you. Okay. So retention becomes paramount. So what ends up happening in a lot of these agencies is-- and I'm just trying to give the audience kind of a, a, a slightly different example, and I want you to maybe add value or poke holes where you see there could be other things in here. But just like I could see a spot where thinking about what you just said, where instead of...

You know, what-- So going back, what happens in these agencies a lot of times is they become heavily service-oriented, and a lot of their human cost and a lot of the cost in general ends up stacking in the service side because they need to retain this business to, to stay profitable. I could see a scenario based on what you just said, where the AI is actually looking at every transaction, looking at every touch point, every text, every email, every phone conversation that comes in and could say, "Hey, you know, this account's like ninety-nine percent guaranteed to retain. Send them this nice pleasant email letting them know the renewal's coming up, but they're really good. Everything's fine. The renewal didn't go up.

You know, they're in a good spot. Good. However, this account over here, here's where you actually wanna deploy your human because this one had kind of a negative text here, and they had a fifteen percent increase in this policy, and we have to rewrite this other policy, and these moving parts can create a lot of issues. And, and actually we've created an email with a calendar link to actually set the appointment and it's waiting for you, and if you like it, just hit go." Like something like that, where now- Yep ... that normal work that a human would have to sort through all these different touch points and probably not even be able to connect all the dots, that could be connected like this, and b- they're-- now they're able to deploy their resources in the specific points where there's trouble and not in the places where maybe, uh, just a kind of classic auto renew with a nice email letting them know everything's fine would do, do well.

Yeah, that's a, that's a great example. And like being able to use that in ways that not only, not only help your retention, but like allow you to deliver service at a level that was like impossible before. Like if you could have an employee that was like dedicated to every single customer, even, even though in your business it would have never made financial sense to do that otherwise, um, now suddenly you can do that because of AI. Um, uh, I'll give an example that's just like what you, what you mentioned about the retention is that, uh, we have, uh, AI support with, with Heights AI in our software, so somebody can ask it questions about, uh, how to find something or can even ask it to do the thing for you. But, uh, we wanna encourage everybody to reach out to the human support.

And we know that when we deliver human support, that we can help the, the creator better, and then they'll probably stick with us longer. And so what happens is a lot of-- you know all the systems that have the things like you have to bug the, the little old chat bot and say, "No, I wanna talk with a person. I don't, I don't want this, this bot." Uh, with our system, with Heights AI, is you can talk with a person anytime you want. You can, uh, you can go and email us anytime you want.

You don't have to go through the AI. But if, uh, Heights AI determines after a conversation that the person had some kind of bad experience or they're having trouble, it will act- actually escalate that on its own to our human team, so that way we can look at it, and that way we can see, oh, this creator may need help with something. Here's what it is. Here's what happened. And then we can actually step in proactively and say, "Hey, uh, it looks like, uh, you're trying to get help with something.

Is there anything else you need?" And now we get to help them at a level where like previously we would have maybe just not even known they were stuck before. Dude, I, I love that example. It's, it's this idea, um... I was talking about it with a friend the other day.

Uh, he's got a, a different type of business. He's in, uh, finance space, and we were talking about seeing around corners, right? And that-- this was-- these were his words, and I, and I love them. Um, he's like, he's like, "It's letting us see around corners that we couldn't have seen around before." And, and I think the example you just gave is perfect.

So much of retention, if you're able to get a postmortem on why someone left, is just you were completely reactive, right? People want, and m- I think more and more consumers, customers, clients want proactive service. They wanna know that you-- It shows that you care when you are willing to reach out before that client has to reach out to you. Like if you can-- That, that is such a powerful touch point to say, "I see that you're struggling here," or, "I see that something's about to happen that could cause a problem for you. Let's figure out a way to solve this problem or, you know, step around this obstacle before you even hit it."

That could be the difference between someone leaving you on that renewal or that next month and that person being a customer or client for the next ten years. Because now they know you give a shit, right? I mean, it, it really shows that you care when you're willing, able to, to, to, to step out front and say, "Look, there's a pothole coming, and I don't want you to step in it," right? "Here's what we need to do." And that type of insight, it's not a failing of humans.

It's a, it's a, it's a... You know, 'cause there are humans that can do that in very specific niche moments. But in a broader sense, as you scale your business, it's impossible for us to manage all those different data points and, and then also project into the future might, might be possible. But the pattern recognition explicitly of AI creates this scenario like you just described that, oh my god, it's just so incredibly powerful. Yeah.

Yeah. So we've kinda level set vibe coding works. We gotta be smart about it. There's way more to vibe coding than just one-shotting something and putting it out. Anytime someone talks about one-shotting, be very weary, right?

Yes, you can get a prototype one-shotting, but commercializing something that you one-shot is not necessarily reality, I would say. Uh, so there's a lot to it, but you can create very viable tools. You can create personal tools. You can create all kinds of stuff, which is great. Let's, let's kinda move towards the future 'cause one of the things that I think, you know, we talked at the beginning, like I'm a huge AI optimist, and part of the counterargument that I get is someone will try to straw man my optimism with what AI can do today.

They'll be like, "Well, today it makes mistakes." Actually, I'll give a great example. I just saw this on X, um, this morning. A woman went on and did, you know, one of those talking head things, uh, where she's like, "You know, I, I looked at my insurance policy," and this just coincidence that this is insurance, but, "I looked at my insurance policy and AI had misclassified my job category, and when I corrected the job category, uh, all of a sudden my premiums went down, you know, 160 bucks a year or whatever," blah, blah, blah, "Be careful of AI." And my response was like, well, I'm 99% sure a human is the reason that, that you were classified wrong because we never t...

We wanna bash AI for one mistake. You know what I mean? Like, like a human can make 100 mistakes, and we're like, "Ah, you know, that's just business." AI makes one mistake, and it's terrible, it doesn't work, and it's never gonna be the future. So I think we, we have to, if we're going to believe in AI and integrate it into our business, we have to kinda future cast n-not just what the reality is today.

So when you're looking at Latch Loop and, and building a, a, a tool that people can use to build for the future, I mean, that's essentially what you're giving people is access to a product that will allow them to build to the future, where do you see this stuff going? Where, what do you see as possible in the near term, we'll say one to three years, um, that maybe today people are missing or j- or maybe isn't as secure or doesn't work as well today but you know will be a problem that is solved, and that's something that people can leverage, um, you know, if they stick with this and they believe and they commit to it? So, uh, it's a great question. I think, uh, what I'm gonna say is something that, that I think is gonna be solved better in the future, so you don't have to understand it as well, but is gonna be something to, uh, if you understand it now, it's gonna help you get so much more value out of AI and un- being able to separate its shortcomings from understanding what it actually is and how it works in the behind the scenes. Um, so like I'm sure we've heard these stories of somebody using like OpenClaw or some kinda agent, and it, it deletes all your email or it does some kinda crazy thing.

And first of all, in your business, if you're using these tools in business, you probably want them set in a way that you're not like relying on hoping the agent does that, but instead you have these e-enforcements in place that it can't actually go and delete all your email or something like that. Um, but what I would tell people is when you see that, uh, AI gets something wrong, and we imagine that, uh, the AI is very much either like the way a human would work or we imagine, I think, the, the sci-fi version of AI, that there's this, this magical thing computing and always thinking and growing in the background. But what we know is the real way that these models work is that they're only essentially like alive for the moment that they're kinda running inference and responding to us from that prompt. And when you think about it that way, I think that this changes the way that maybe you interact with it because when we put the agent in a loop, and it continues to work on something, the way I would describe this, like metaphorically, is that we're actually, uh, having this AI kinda come to life and saying, "Here's all this information. You gotta do something with it."

And the AI does have some kind of sense to know that it's gonna basically exist for the next couple minutes, that it has to respond with something about that. And so what the AI's doing in its training is it's trying its best to do whatever you said in the next minute and deliver something. It may be what you would determine is actually half done or actually incorrect, but if you can realize that that's what the AI's trying to do, and then after that, if you put it in a loop, it's technically another AI. It might... You might be feeding it the context of what happened, but i-in a way it's another AI.

So it's like being brought to life over and over again of all these different agents with, uh, these different memories that you're kinda forcing into them rather than one thing that's always working. And so if you think of it that way, I think you can start to think about how are you giving it the right information so that way it can perform the task that you're looking for. I think that's a really important point for people to understand. And, and I actually-- So I have an OpenClaw that I play with. Um, his name's Maximum Effort, Max for short.

And, uh, what's funny about these things, and I, I said this the other day on the show, I was like, "I can see why there's all these men who are like forming these emotional and relationships with this thing because, one, unless you explicitly tell it not to be, they tend to be very sycophantic. Two, conversationally, when it's, when, when you've provided, uh, especially like an OpenClaw with the right, like soul.md file, identity, when it starts to understand who you are and how you like to be responded to, it can feel like a very real relationship. And as a kind of a test and just I was interested in its response, I said like, "Are you alive?" And what was really interesting was it came back, and it said, its first answer was, "No, I'm not alive." It said, "But if, if I were to personify what I do," it's exactly what you said.

It goes, "I'm not like hanging out with other AI." This is literally what he said. He's like, "I'm not like hanging out with other AIs like on the internet when I'm not talking to you." He's like, "I essentially don't exist unless an, unless a cron job or an automation is running in the background, and I have something I have to do, or you're conversing with me." So I think, I think that while maybe logically very obvious when people hear it kind of said explicitly, um, I think we can get lost in this idea that these things are just like working all the time, constantly on, searching for, you know, some way to like take over the world and, you know, embody some robot with a machine gun or whatever, like the world's gonna come to an end.

It, it's just not the way that it works. Like it, it has to be told to do these things. And What I like about this idea of looping that, that you're kind of, uh, uh, uh, taking the banner of, and, and I, and I really think it's a wonderful idea, is it's what you're-- what I hear you saying is the first loop is maybe one sub-agent of an AI runs and tries its best, and it delivers that package and its learnings to the, to a second sub-agent that then spins up unique. It's like handing it to another team member, and that agent goes, "Okay, I see what you did here, but you missed this bug, and this isn't fully functional. Okay, I'm gonna fix that.

Okay, great." And then it shuts down, and it hand passes it to maybe a, another AI that then goes, "Okay, I see what you two did here, and I see that bug fix. Okay, that's great, but we're still missing, you know, this connection." And, and now each one is kinda learning from the next, and that saves you from having to know what those things are because the AI's able to learn from the next one versus you having to go, "Okay, what did it do here? Now, what would I want the next step to be?" 'Cause I know in the very early coding agent tools, that was where I started to get lost.

I was like, "I just don't know what I don't know." Like, I, I don't know what the next question is to ask 'cause looping wasn't a thing for, you know, in twenty twenty-- early twenty twenty-five when I first started. Like, that wasn't even something you could get these tools to do. Yeah. Well, it doesn't even have to be as complicated as, like, having to technically set up some kind of sub-agent or something like that.

Um, what it's about, again, is, like, directing the attention. And so if you're building the software or building some kind of thing, then you can say, like, "Okay, this is the thing we're building," and then maybe the, the next equivalent, uh, agent, you don't have to say anything about agent. You're just defining it in, like, a long prompt where it's like, okay, then we need to, to check over all this for security, or we need to check over this for optimization, or we need to go and do some research on the web to make sure this aligns with the, the marketing thing that we're trying to do. And so it's distributing where are the things that we think are important to kinda spend some attention. Yeah.

Do you think, in general, people don't use long enough prompts when they start to build? I think it, it comes back to, again, like, just building that sense of actually trying these things out and working with the models because, uh, uh, I saw-- When I saw everyone talk about, uh, looping, I saw the founder of OpenClaus says, "You should be working on building these loops. These are, this is the future." Um, I would actually push back on that a little bit and say that it's not just about building loops because if we have the infinite loops for everything, then everybody just has a slop factory, right? And so we have to realize, like, where are the parts that we wanna have a back and forth where we're iterating on something that we care about, and we need to see, okay, what does the interaction here look like, or we need to see some kind of information before we can tell the agent to continue versus something where we're able to articulate, like, this is a very clear thing that needs to be done, and the, the minor specifics of how something needs to be accomplished is not so important.

That's the kind of thing like, okay, the agent can just be working on this in the background. And yeah, so it's kinda separating and getting the, the skill for yourself of learning where's the thing that the agent can just be working on for me versus the thing that I need to put more of my attention to. Yeah. I, I love that you've brought up multiple times now this idea of allowing the AI to interview you to get to a better answer, and my own experience with that is, um, I just signed, uh, my first book deal, um, and how I got there, uh, 'cause I had this idea for the book. Um, I have this, uh...

I do a lot of leadership and growth coaching. That's basically my career, either as a CMO, CEO, or now as a, as a coach and a consultant. Um, and I've always-- I had, I developed this concept on how I train people to get the most out of their people, which is called a human optimized business model, which clearly defines what I'm trying to do but is a non, not a very good brandable name for something. So I call it Easy Mode, is what I call it. Okay.

And but, uh, so I had this idea, and I had these experiences of, of whatever, uh, uh, of, of doing this in multiple businesses and training people and implementing it in my own businesses, et cetera. But, like, it's not like it was one coherent thought. So I had a Friday where, uh, I'm divorced, and, uh, so my kids were with their mom, and, uh, the woman I'm seeing, she was out of town with her kids, so I'm all alone. It's a Friday night, and because, you know, I'm forty-five and at this point, you know, kind of a nerd and a loser, um, I just told, I said to my OpenClaw Max, I said, "Hey, like, I wanna develop this idea. Here's the core concept.

I want you to act as," um, I gave him a couple different, like, versions of this, but I said, "One, as, like, a leader who I'm training, one as a book publisher, what they would wanna see, two as one of the greatest ghostwr- or three as, uh, you know, one of the greatest ghostwriters ever." Yeah, I'm kinda broad-stroking what I said, but that's kinda the core idea. And I said, "I want you to interview me." Dude, it was six hours. I sat there, and this thing just-- And, and, and my-- It doesn't have to be six hours, guys, so I'm not saying this is always six hours.

I allowed it to go six hours 'cause I was so fascinated by the process, and what I thought was amazing was because I gave it a pers- a personality to act as and some guardrails as to where we were trying to go, it just kept digging in. Like, I would share a story, and it would go, "Well, what happened in between this part and this part?" And I was like, "Oh, shit." I had never really, like, thought about or explained or, or verbalized, like, what happened between those two parts. "Okay, well, here's what happens in those two," blah, blah, blah, blah.

And it would go, "Yeah, but, like, that's too broad. Like, give me a specific..." "Okay, well, let me think of a sp- okay. Well, back in twenty twenty, I was talking to blah, blah, blah. Here's what we did."

And it forced me to think about the, the core idea of, of Easy Mode, this, this idea that I'm writing the book about, like, at a depth that I don't-- I would've never gone. If it was just me, like, if, if this didn't exist- I don't even know if a human could have interviewed me to the depth that this took me and the amount of specific experience and ideas, and then it would even came back a couple times and said, "Well, this is actually conflicting ideas. You said this here, and then this idea kinda like whi-which, which one is what you mean, right?" And then I was like, "Oh, shit, like I, I didn't even realize those were conflicting ideas. I actually kinda forgot an hour ago that I say said that thing to you.

That's really interesting. Okay, actually, the original version is right, and I didn't really mean to say it that way." And, and my point is, like, for what-- for trying to accomplish goals, especially really important goals or large goals and this idea of thinking big, right? You can give, "Hey, I wanna think bigger about this idea, act as this thing, and interview me until you feel that we've satisfactorily, satisfactorily, uh, uh, developed this concept into a big idea." I don't know that there's a better tool out there for extracting information out of your own mind than being interviewed by AI.

Yeah. Yeah, I think it's such a great, uh, a great way to work with it because, um, it's not like I, I'm not gonna be more successful with, with, uh, using some kinda agent because I'm a better writer or something. It's, it's actually just because of the process of figuring out, like you said, there might be things that you forgot to even mention that, oh, well, this is important. I should say I do have a very strong thought of, like, the way I want this to go, but I didn't mention that. And so getting AI to, to figure that out, um, so that way whatever you tell it, it has the things you actually care about. 'Cause so many people I see say, "Okay, oh, AI didn't do what I want.

Look at this thing. It's clearly bad here." Well, did you tell it? Did you, did you say what you cared about in that, in that instance? And so if you can, uh, be able to communicate those things ahead of time because the AI helped ask you, um, then you'll end up with a much better result.

Brian, dude, I could talk to you all day about this stuff. Um, love to have you back on again in the future as you develop and as these things start to change. This, these are some of my favorite conversations. It's just, it's like the Wild West, and to me, the people that I see thriving right now are those who embrace the fact that this is the Wild West to a certain extent, and that you, you-- there is no right or wrong right now. Everyone is, you know, even the most sophisticated users.

I mean, I heard Chamath Palihapitiya, who is one of the smartest businessmen who understands, seems to have a really good perspective. He has AI businesses. He's on the All-In, not to mention the All-In podcast. I seem to be promoting them today. I'm not.

Uh, I don't mean to be. Um, you know, he even will say things on that show where you can tell, like, he just, he just hasn't made up his mind yet. We just don't know where this is going. We may have ideas. We may have some thoughts or, you know, past experience we can pull on, but I think it's fair to say that no one knows exactly where things are going, and the only way to get there is to, is to play around, figure it out, test, build.

And tools like Latch Loop, uh, are a wonderful way to get started in a, in a constructive and defined way where you don't have to use a terminal and Claude code on your computer if that makes you uncomfortable. So that all being said, um, where can people go to learn more about Latch Loop, about Heights Platform, connect with you? Where should people go to go deeper into your world? Yeah, thanks. Uh, great talking with you.

I completely agree. Um, this is, uh... You gotta just build things. Nobody knows what they're talking about. This is all, this is all brand new, okay?

Who's to say that, that anyone who's built any kind of agent product, that that's really the best way to go going forward? So, like, what's so incredible is, like, whoever is listening to this right now, like, they can go out and potentially build something better than OpenAI or Anthropic or whatever in terms of the actual agent and the workflow and how things are going. Um, if you wanna check out what I built, uh, Latch Loop is available at latchloop.com. We have a free trial. We're giving free, uh, GPT 5.5 credits for that.

Um, uh, Heights Platform is heightsplatform.com, and, uh, that also has a free trial, no credit card required. Um, yeah. And any, any socials, any place where someone can follow along with you- Socials, um- ... as you're creating content? I'm not, I'm not super active on socials. I'm on X, @BrianMcNulty.

Awesome. And, uh, I also, if you're interested in, like, the creator space, I have my own podcast, uh, called The Creator's Venture. Tremendous. Guys, we'll have everything linked up, whether you're watching on YouTube, wherever you listen, just scroll down. You'll find all the links.

Brian, dude, appreciate your time, man. This is a phenomenal conversation, and I love that you were willing to kinda go everywhere from, from basic aspects of this all the way to, to some more advanced concepts. I think it's really important to give people, um, kind of the full spectrum of what we're talking about. Thanks so much, man.