Brian Balfour: Why ChatGPT is the next big growth channel

Lenny Rachitsky 0:00 Everyone's always complaining, SEO's dead, it can't grow, word of mouth is so hard.

Brian Balfour 0:03 All of the ingredients for a new distribution platform are essentially happening. My prediction, the new distribution platform will be ChatGBT. There's a bunch of signals that they're about to launch that.

Lenny Rachitsky 0:14 This is a huge opportunity for companies to get on it.

Brian Balfour 0:16 It ends up being a prisoner's dilemma. Don't trick yourself into thinking that you can't play the game. The cycles seem to be getting shorter and shorter, so you actually have a smaller amount of time. If you don't do it, your competitors are gonna go to the new platform, and your customer expectations change. There is no opting out of the game.

Lenny Rachitsky 0:31 This is the opportunity to disrupt an incumbent.

Brian Balfour 0:33 If you're a late stage company, you place multiple bets. For startups, it's a totally different ballgame. You have to choose one and go all in.

Lenny Rachitsky 0:40 Think about companies like Zynga that grew on Facebook and then became massive companies.

Brian Balfour 0:43 Building a great product is one of those things that's necessary but not sufficient, and actually the separation is between those that build really great distribution.

Lenny Rachitsky 0:52 What would be the backup, if not ChatGBT?

Brian Balfour 0:54 My hypothesis of who's best positioned would actually be... (music)

Lenny Rachitsky 0:59 Today my guest is Bryan Balfour. Bryan is the founder and CEO of Reforge, a company that I've been a longtime fan and advocate of. Historically, Reforge is focused primarily on teaching courses on product and growth. But more recently, they've transitioned to building their own products, including a product called Reforge Insights and a bunch more really cool stuff coming very soon. Prior to Reforge, Bryan led growth at HubSpot. And over the course of his career, he has seen the rise and fall of every major distribution channel, including Facebook's ad platform, Google Ads and SEO, and the Apple App Store. Based on what he's seeing, he is predicting the emergence of a brand new and powerful distribution channel that will likely arise in the next six months, centered most likely around ChatGBT. It is really rare for a new growth channel to open up. It's been a long time since the last one appeared, and the people who recognize this and hop on it early are the ones that reap the most rewards. So this is a huge deal. In this conversation, Bryan shares what he's predicting, what he's seeing, why this is a big deal, and what you should be doing about it right now. I highly recommend you listen to this full conversation and discuss the ramifications with your team. If you enjoy this podcast, don't forget to subscribe and follow it in your favorite podcasting app or YouTube. Also, if you become an annual subscriber of my newsletter, you get a bunch of incredible products for free for one year, including Lovable, Replic, Bolt, n8n, Linear, Superhuman, Descript, Whisperflow, Gamma, Perplexity, Warp, Granola, Magic Patterns, Raycast, ChatBRD, and Mobbin. Check it out at lennysnewsletter.com and click Product Pass. With that, I bring you Bryan Balfour. Today's episode is brought to you by DX, the developer intelligence platform designed by leading researchers. To thrive in the AI era, organizations need to adapt quickly, but many organization leaders struggle to answer pressing questions like which tools are working? How are they being used? What's actually driving value? DX provides the data and insights that leaders need to navigate this shift. With DX, companies like Dropbox, Booking.com, Adyen, and Intercom get a deep understanding of how AI is providing value to their developers and what impact AI is having on engineering productivity. To learn more, visit DX's website at getdx.com/lenny. That's getdx.com/lenny. This episode is brought to you by Basecamp. Basecamp is the famously straightforward project management system from 37signals. Most project management systems are either inadequate or frustratingly complex. But Basecamp is refreshingly clear. It's simple to get started, easy to organize, and Basecamp's visual tools help you see exactly what everyone is working on and how all work is progressing. Keep all your files and conversations about projects directly connected to the projects themselves so that you always know where stuff is and you're not constantly switching contexts. Running a business is hard. Managing your projects should be easy. I've been a longtime fan of what 37Signals has been up to, and I'm really excited to be sharing this with you. Sign up for a free account at basecamp.com/lenny. Get somewhere with Basecamp. (music) Bryan, thank you so much for being here, and welcome back to the podcast.

Brian Balfour 4:09 Yeah, thanks for having me. Excited for this one.

Lenny Rachitsky 4:12 I'm really excited to have you

Lenny Rachitsky 4:13 back. Uh, we're just gonna dive right in. Essentially, you've uncovered a, a really important trend, uh, or insight about how products are gonna grow differently in the future, how growth is changing, and this is something that I think a lot of people need to hear, so I asked you to come on to share what you're seeing. Uh, I also think this is just very timely. Uh, uh, I think you said, like, in, you're gonna say in the next like six months things might significantly change. So I'm really excited to do this. We're gonna spend this whole conversation on this, on this insight. To set us up, what is just the big idea? What's the high level idea here?

Brian Balfour 4:46 Just like you, I've spent my whole career just like really passionate about startups, you know, figuring out how to build products that win, that emerge in new markets, and, uh, one of the things that I have, you know, learned over time, or one of the things you hear a lot is, from a lot of folks is, uh, to win you have to, like, really build a great product. A lot of the advice kind of boils down to that.

Brian Balfour 5:09 And one of the things that I feel like I've banged my head against the wall in a lot of ways in my career is actually telling people that building a great product is one of those things that's necessary, um, but not sufficient, and actually the separation is between those that build really great distribution. And so, uh, this general partner, his name's Alex Rampell, he's, uh, at Andreessen Horowitz, actually wrote this blog post 10 years ago, um, uh, back in like 2, like 2015. In the essence of the blog post, he basically says, uh, one thing, which is that startups is a game of trying to get distribution before the incumbent can copy, right? So it's this kind of concept of escape velocity. And, and so, you know, on that, uh, note, right, which I think is, like, a very good summary of, like, what you're trying to do in a startup in distribution is that...We're right now living in this environment where that game of startups kind of getting distribution faster than the incumbent has gotten (laughs) way harder in a lot of ways and in some small cases has gotten a little bit easier. But if we think about this, the way that it's gotten harder and some of the things that probably a lot of founders or- or folks working on the growth side have probably feel is that, one is that incumbents can copy faster these days, right? So that window that you have to get that escape velocity has actually shrunk. It's, uh, it's decreased. The second thing is that the... a lot of the organic distribution that we've had, especially over the past few years, has really shrunk as well. So everybody's talking about the decline of SEO and, you know, clicks declining, but you also see it in some other cases, right? A lot of these social platforms don't really let you send as much traffic to sites. You know, LinkedIn just changed their algorithm, which has really dropped organic distribution. Obviously, the- the Twitter to X transition, that happened, right? Like, TikTok's almost always been like that. And then the third way that it's gotten harder is that AI is really good at writing software, right? And code generation. And so everybody's kind of feeling this infinite increase of competition, especially at the startup level. And, you know, YC's pumping out six of the same thing every single cohort, right? (laughs) Like that's what- that's what it literally feels like. So it's gotten way harder. This game, this escape velocity game has gotten a lot harder. It's gotten easier in some very exceptional cases, like a Cursor or something, where AI's kind of been like the spark. You know? I know you wrote, uh, the blog post about the race car engine, and I think you said, like, uh, there's like the spark plug in the engine, right? And- and so AI kind of really created that, uh, a- a new type of spark, a new type of interest of early adopters to- to- to fuel some- some new players in a short period of time, right? And so it's amazing to see something like Cursor overtake market share of something like GitHub Copilot in nine months or less, right? Like that- that's how fast it happened. It's kind of crazy. But the main thing that people need to understand is, okay, well if that's the game I'm playing, right? How to get to escape velocity before the incumbent? Like what are all the ways to do that and, um, and to really figure that out. And there's multiple ways that this can

Brian Balfour 8:14 happen. But one of the major ways, one of the major, major ways that we always see is that, uh, this can happen when new distribution platforms emerge. Um, and 'cause when new distribution platforms emerge, startups are usually the fastest to take advantage of them. It's slower for the incumbents to move. It gives startups this opportunity essentially to play this game. So Casey Winters wrote this blog post about two years ago, um, maybe like 18 months ago, about, uh, the AI technology shift. And his key point was the AI technology shift has been a technology shift that has not come with a distribution shift yet. So if you look historically, we've had a bunch of technology shifts from, you know, uh, the t- internet to the cloud to mobile to social, like all of these different types of things. And some of them come with distribution, new distribution platforms, new ways to dis- distribute products and some of them don't. But the most powe- powerful ones, the most impactful ones are the ones that, um, do come with these new distribution platforms. And so, uh, his second key point was that these two things don't actually happen at once. Usually you get the technology shift, then you get the distribution shift a little bit later. So now we're a couple years from that post. We are a couple years into AI technology shift, and one of the things that I am seeing is all of the conditions, all of the ingredients for a new distribution platform to emerge are essentially happening. And so I think we're at an inflection point where we're going to see this emerge really fast.

Brian Balfour 9:45 And the key thing for everybody to know is that as new distribution platforms emerge, they follow the same four-step cycle, and it's kind of a game that you're playing, that everybody's playing. And so just like any game, you kind of need to know the rules of the game. You need to know the steps of the game in order to have any sort of opportunity to win. And that's kind of like the- the- the thing that, uh, that- that I've lived through once again, both painfully and, um, also in good ways, and it's something that I'm keeping my eye on and something that I've been talking about. So before we go into that, you know, four-step cycle, I figured I'll- I'll pause there to see if you- you have any follow-up questions on that.

Lenny Rachitsky 10:24 Okay. This is amazing. So essentially what you're saying is we've all these ways to grow. There's li- uh, SEO. There's paid growth. There's sales. All these, uh, channels have been around for a long time. They're extremely saturated. Everyone's always complaining SEO is dead. It can't grow with SEO anymore. It can't grow... Word of mouth is so hard. There's so many amazing things now. That's hard. Uh, paid is so hard. It's just, uh, like all this money just-

Brian Balfour 10:47 Yeah, packs are rising. Yeah.

Lenny Rachitsky 10:48 Exactly.

Brian Balfour 10:49 All these things. Yeah.

Lenny Rachitsky 10:50 So all these saturated channels, and y- what you're saying is there's an emerging new channel that has not yet been saturated, and this is a huge opportunity for companies to get on it. Uh, and you'll talk about timing 'cause it's a little tricky to even know exactly when to go big on this.

Brian Balfour 11:05 That's right. Yeah.

Lenny Rachitsky 11:06 Um, but that's a huge deal. This has been a long time (laughs) since there's a new way to grow that you can actually use as a-

Brian Balfour 11:11 Yeah.

Lenny Rachitsky 11:11 ... lever for growth and not just hope for the best.

Brian Balfour 11:14 Yeah.

Lenny Rachitsky 11:15 Okay.

Brian Balfour 11:15 That's right.

Lenny Rachitsky 11:15 Uh, before you get into the- the cycles, do you wanna tease what the answer is just to give people a little hint? Or do you wanna keep it secret?

Brian Balfour 11:22 (laughs) Well, to be clear, right? Like, okay, so, um, my prediction, well, we don't- we don't have a clear winner yet. Um, my prediction of the new distribution platform will be ChatGPT, um, in some ways that people probably already think it's happening and some ways that it won't. Uh, but the thing that is less important or, like, that is more important than whether I have predicted the exact winner correctly, the thing that's more important is to understand the cycle and- and evaluate like how to determine where you want to place your bets and how to place those bets, which I know- which I know we'll talk about 'cause I- I'm... I could be wrong about the ChatGPT prediction and what's gonna happen there. Um, there's... I think there's gonna be two parts of it. There's going to be what they do with like a ChatGPT search experience, but I think the bigger thing will be whatever they do with launching a third-party platform on top of Cha- ChatGPT. There's a bunch of signals that, um, they're about to, uh, that they're about to launch that.I'm pretty sure it's going to be ChatGPT. The thing I'm way more sure about is that some new distribution platform will emerge and it will follow the same four-step cycle. That, that, that's kind of, that's kind of the key. So could be wrong on the first piece, I am very confident on the second piece.

Lenny Rachitsky 12:40 Okay, excellent foreshadowing. Uh, I completely agree. If it's anything, it would be ChatGPT at this point. Let's get into it. What are the, what are the cycles that, uh, platforms generally follow?

Brian Balfour 12:49 Yeah, and I'll give some examples of this, but let me explain the four, uh, the first four step, uh, the four steps of the cycle first, and then we'll go through a bunch of examples of all, all those individual steps. So the, the four steps are essentially, one is like I call a step zero. It's, it's the, the conditions of the market have been met. Um, step one is about a moat, step two is about a platform opening, and step three is about, uh, the platform closing for control and monetization. So let me kind of briefly explain each one. Um, step zero is about the competitive market being, uh, met, uh, the conditions being met. And there's a few part piece of this. One is that typically what happens is that a, uh, there is consensus that there is going to be this new huge category, right? Think social, think mobile, like all those types of things. And in this case, right, uh, these AI like chat platforms, like a ChatGPT or a Claude. So there's consensus about that, but there's no clear winner yet. And you, you typically have somewhere between five to seven, you know, major players really battling it out, right? So, and they're all kind of looking for what is the edge, what is the thing that is gonna, that is going to, to help me win because all of these dynamics you... In, in all the history, they either end up in monopolies or duopolies, right? And so the, the stakes are really large, and so the competition is fierce. So that's kind of step zero. And I, I think we could all agree that we are in that mode (laughs) you know, right now. We've got, we've got OpenAI battling with Claude, battling with Gemini, uh, and Google with whatever Meta comes out with new, their new team, you know, so on and so forth. Like, and there's huge amounts of capital, there's consensus, like all that type, they are in a fierce competition. So that's step zero. Step one is then, um, these players, somebody essentially identifies whatever the moat is, the thing that is going to help build them defensibility and help them hit escape velocity and become that monopoly or duopoly in that, uh, single category. And once they figure out what that moat is, then they need to press the advantage, right? They need to figure out how to gather that moat as fast as humanly possible. And it tends to be that you can't do that by yourself. And so you kind of need the help of an ecosystem in order to gather more of that moat, and that typically comes down to third party content creators or app developers and other businesses. And so they all establish like a third party platform, right, that has some incentives built in. And usually the value exchange is, "Hey, you develop on top of my platform, right? You add more use cases, you know, more engagement, like all, all of these things to my platform. And in exchange, I'm gonna give you something in return." And usually that thing that's in exchange is, "I'm gonna give you some new form of distribution for your application and for your business." But what essentially happens is we go, uh, over time is that we go into step three, which is the closing period, which is at some point all of these companies end up, um, starting to lock down the platform. And this tends to happen for reasons of like monetization and growth, right? Uh, they either competitively don't want, you know, somebody to use their own, you know, platform to disrupt themselves. Like we saw that in the early Twitter days with things like Vine and Periscope, right? Like shutting those things down unceremoniously, right? Or they need to find ways to monetize at a deeper and deeper level because all these companies, like they have to grow. And, you know, Google's the classic example here of just more and more real estate has either been taken up by either ads or their own, you know, first party applications. And so that's the key is like they close it down by doing one of two, by one of a few things. They either shut it down entirely; two, they develop their own first party applications to absorb the, the highest use cases; or three, they artificially depress the organic distribution that they gave you in the step prior, uh, to push you towards paid mechanisms in order to monetize. And so I think we should go through like multiple examples here-

Lenny Rachitsky 17:07 Mm-hmm, mm-hmm.

Brian Balfour 17:07 ... but that's kind of like the core essence of, of the four steps. And so I'll po- I'll pause there.

Lenny Rachitsky 17:12 Awesome. So it's essentially figure out what's gonna make, create defensibility long term. What's your moat? Bring everyone in. Hey everyone, welcome to, to Facebook. Uh, everyone joins Facebook and then okay, uh, all the developers build on Facebook to bring in more people on Facebook. And then they're like, "Okay, now you gotta pay. There's a toll," and, but you love this so much and you're so hooked, all your friends are here, you may as well stick around.

Brian Balfour 17:34 That's right. That's right.

Lenny Rachitsky 17:35 Uh, amazing. Okay. So yeah, a few examples would be great.

Brian Balfour 17:38 Yeah. So you, you just hit on the first one and the fir- this is the first one that I always think about because this is where I learned about this cycle very early in my career. Like my, one of my first companies, uh, was during the whole, uh, uh, the Facebook platform boom, you know, social gaming, all of those applications. And I lived the full cycle in a very short period. I lived the, the glory days and the ho- just the absolute horror days, and the, and it was very painful. But this is exactly what happened. So let's go through the four steps. So step zero, Facebook was in a brutal battle with, uh, MySpace, Friendster, um, and a few others. People forget this. People forget that there was actually like a bunch of competitors at that time. And in fact, those competitors were bigger than Facebook at the time. They had more users back in 2007 when Facebook launched, um, their third party platform. Uh, but one of the key things is that Facebook was very early to the insight about-... uh, the direct network effects in that, um, there's gonna create real lock-in that the, the more friends, the more of the global network that was on there, the more that it was just gonna feed and, and hit this escape velocity. And so at the time they launched their platform, I think they were maybe like one-fourth, one-fifth, uh, the, the size of, you know, something like Ma- MySpace or even Friendster, Orkut. Like some, these are some of the names at that time. But they opened up their third-party platform. And what was the value exchange? They went to third-party developers and they said, "Okay, we've created this canvas." Um, they used to call it the canvas. And they were like, "You can put anything in the canvas that you want; an app, a game, whatever. You can monetize in any way you want. We just want this sidebar real estate on, on the ads. That's, that's what we're really interested in." And so there was this ma- mad gold rush o- on, uh, on that Facebook. I'm sorry, the other part of that was not only will you put it there, we're gonna give you access to all of these notification channels and feed to get distribution for your application. That was, that was the other piece of it. And so you had this mad rush of developers, um, coming in and you had this huge, uh, like social application, social gaming boom. Uh, people just grew incredibly virally, uh, very fast. But eventually, essentially what happened over time is they kept peeling back that value exchange. They first were like, "Uh, actually, you know, that, those dollars that you're, you're making inside that canvas area, well, we want a percentage of that," right? (laughs) And so they changed that. And then they figured out their a- like, their ad systems and then they started peeling back. You know, they started suppressing access to all of the organic channels that they had. Eventually they went all the way towards absorbing the highest, uh, you know, use case into their own first-party platform. Things like, uh, first-party applications, things like events, photos, all those types of things, and basically shut down the platform, um, for dead. And these companies that had basically built on top of this platform, you know, uh, the other thing is by the time they started closing all those things down, all those competitors that we talked about, they were so far ahead at that point because they had built off the back of all these developers coming, adding use cases, bringing more users onto the platform. Like all, like identifying that moat. They were so far ahead it didn't matter. It didn't matter what the other, what the other folks, uh, uh, did at that point. And, and that's what kind of really gives you, uh, confidence to start closing down. But there's so many other examples of this, right? If, if we go through it, right? So-

Lenny Rachitsky 21:02 Just, before you get other examples-

Brian Balfour 21:03 ... yeah.

Lenny Rachitsky 21:03 ... just something that I'll highlight here. One is the moat they, they identified in theory was the friend graph, I imagine. Just like once-

Brian Balfour 21:10 Yeah.

Lenny Rachitsky 21:10 ... we have all your friends, you're not gonna want to go anywhere. I imagine it's also, uh, important to note, you may not... This is kind of like a natural thing that would happen if you build the thing and it grows and you're like, "Oh, maybe we should change strategy." I imagine not everyone even knows this is what will happen and they kind of organically evolve their strategy. Or do you think everyone's just like, "This is are, gonna be our plan, step one, two, three, four"?

Brian Balfour 21:32 I think a different version of that question is I think some people could sit here and interpret this as all these folks are evil (laughs) , right? And, and that's not what I'm saying, right? Like, that's actually not what I'm saying. I wanna be very, I wanna be very clear on that. Uh, because I think, you know, a lot of the, a lot... This cycle happens because of competitive and capitalistic, like, dynamics and pressures. It's, it's the same environment that enables, like, creating amazing new companies, right? Here in the US. So... And, and there's, there's like two sides of the coin. And so you go through this cycle because it's a competitive environment. You're trying to figure out how to beat competitors and this is one of the strategies to beat competitors. But at some point, like, you just have to continue growing. You still have, you have to grow those dollars. Like, uh, the, the market does not reward flat companies (laughs) if anybody's noticed. Like, you have to keep growing and so they have to keep finding ways to grow as well as prevent their own disruption, right? They, they can get so big and they can give access, so much access just of distribution to new developers. They don't want to enable, uh, their own disruption as well as, like, that they need to keep growing. And so, uh, my guess is anybody who is kind of sitting in their shoes, you know, owning their platform, is gonna follow the exact same playbook and the, and, and the exact same reasoning. And, you know, look, sometimes it happens also because it actually is the best thing for the user. Facebook's channels did get super spammy and, like, a- all of those things. And that was part of the reason. You know, they, they play this, but let's be honest, it wasn't the only reason, right? Like, a lot of it was, a lot of it for, was, was for these other reasons. And so I don't think it's evil, it's just you need to know how to play the... You just need to know how to play the game. That, that's competition. That's business. They're playing you, so you need to play them. Like that, that (laughs) that might be a little, little sadistic or something, but, but that is, that is business. That's, you're, you're in a game of competition. Yeah.

Lenny Rachitsky 23:23 Essentially, the incentives are pointing you in this direction. Capitalism, this is how capitalism works.

Brian Balfour 23:28 Yeah.

Lenny Rachitsky 23:28 Um, and so it'll pull everyone in this direction even if they, they do- they want to avoid it. Let's do a couple more examples.

Brian Balfour 23:33 Yeah, we'll go through them quick. You know, I, I think everybody's probably... Google's an interesting one 'cause it played out, uh, uh, over a much longer period of time. You know, Facebook happened over the course of about, mm, five-ish years. Something like that. Google kind of did it, like, very slowly over years, but, uh, same thing, right? Massive, early massive competition against Yahoo! I don't know, Alta Vista. I, like, 'cause you name them all, right? Uh, that was even before my, that, that was even before my time, right? They were first to really identify these data moats and incentivizing, um, essentially web developers, content folks to optimize, you know, for their search algorithms and to create this great, great distribution mechanism. Everybody's kind of building content and everything, you know, for them. But o- over time, slowly but surely, right, they did, uh, two things. One is more and more of that real estate, uh-... uh, became ads that they were monetizing. So, they're suppressing organic distribution in order to push people towards the ads, as well as ab- absorbing a bunch of, of the highest value first, uh, party use cases. Things like travel, you know, a- as an example. Or, um, even, like, you know, like, restaurant search and, like, all those types of things. You know, the Yelp, uh, former Yelp CEO and founder is, uh, has been, you know, been out there saying a lot of, a lot of things about, about these practices. So, sa- same exact cycle. Mobile went through the exact same cycle, right? iOS created a new distribution mechanism. They were in massive competi-... They, they had a ton of competition among different phones when they first started on. They found that defensibility was more about the apps, they added all the developers, created the, the app store, like all these types of things. But over time, we've seen more and more restrictions there on that front. And then most recently, right, like we've seen this happen in smaller places too. LinkedIn as an example, first went through this wave with company pages, right? They were like, "Ah, companies, you know, come on, promote your company page, bring in more users," like all that type of stuff. And then y- get, get all these followers. And then of course they, you get almost no distribution now through your company page, uh, because they're pushing you towards ads. And then they recently just did this with personal profiles too, which is they really boosted, um, distribution for individuals to create content for that platform. They then introduced the thought leader ad format, a way to monetize those in- those individual posts, and now you've seen them really pull back on that organic distribution. So, this happens in big forms and it happens, um, even in, in smaller use cases as well. But once again, the steps of the cycle are exactly the same. And the key part about this too is it, the broad trend is that the cycles seem to be getting shorter and shorter and shorter and shorter, so you actually have a smaller amount of time to play the game.

Lenny Rachitsky 26:11 Okay, and the big aha here is, yes, this will end maybe not great for you, but there's this magical period when they're open to-

Brian Balfour 26:21 Yes.

Lenny Rachitsky 26:22 ... customers and users where you can grow like crazy because they want everyone to come and they give you distribution. And what you're saying essentially is ChatGPT, potentially some other platform maybe, uh, is about to enter this mode.

Brian Balfour 26:35 Yeah. Well, let me first, before we get to ChatGPT-

Lenny Rachitsky 26:38 Okay.

Brian Balfour 26:38 ... I think the natural reaction when, when you first realize this, uh, is, "Screw them, I'm not playing that game."

Lenny Rachitsky 26:45 (laughs)

Brian Balfour 26:45 (laughs) Right? Like, that, that, that's what I feel like most people, uh, like, how, how they react, right? Because, um, because the unfortunate truth is, is that a lot of companies don't predict that last stage and end up in a really hard, hard position, right? So many companies got completely killed during, uh, you know, th- the crash of the, the Facebook social platform. Apple's 30% tax, you know, basically destroyed a bunch of types of, um, y- uh, types of applications and business models, uh, because it just, you couldn't, you, like, it wa- just wasn't like margin effective. Like, a- all, so many companies built on, you know, SEO loops, right, that are in serious, serious, uh, trouble right now if that's their only ch- channel. So, so all these things, right? And so I think the natural reaction is, yeah, like, like, "Why would I play this game if I'm a startup or, or a company?" Right? And y- you can even see this with like ChatGPT as an example, right? Uh, they just launched these like deep research connectors. Um, one of them was my former company, HubSpot. And, you know, you, if you sat inside HubSpot and you were just thinking in isolation, you'd be like, "Well, why would I want to make all of my data, um, ac- accessible through ChatGPT and have, like, all of the usage, y- start to accrue there?" Right? Like, it doesn't really make sense in isolation, but we don't, we don't operate in isolation. Once again, we operate in a competitive environment. And what's gonna happen is that if you don't do it, your competitors are gonna certainly go to the new platform and your customer expectations change, and you have to rise to those customers' expectations. Like, they're gonna start expecting you to be in these new experiences, you know, all these things. And so it ends up being a prisoner's dilemma, right? Which is like, you, there is no opting out of the game. You have to play the game. And, and so it's better to be early-

Lenny Rachitsky 28:36 (laughs)

Brian Balfour 28:37 ... than, than to be super late, uh, to this game, especially, especially if you are a startup, right? Uh, that, that's kind of like the, th- the key, uh, that's the, that's kind of like the key opportunity. And so we'll talk a little bit more about how to play the game more, but it's, it's, it's better to be early as well as then the, then the key, the harder part about it is, is anticipating that last stage of the cycle and figuring out how to sequence away from something before that last cycle comes. So, I think that's the key part. Um, but let me pause there and then I'll talk a little bit about ChatGPT and, and some of my reasoning behind that.

Lenny Rachitsky 29:12 Cool. So what you're saying is not only is there going to be this big opportunity to grow. If you don't take advantage of it, somebody in your space will.

Brian Balfour 29:20 Exactly.

Lenny Rachitsky 29:20 So it's not only there's an opportunity, but this is something you need to do because you might miss the boat. And I think about companies like Zynga that grew on Facebook and then became massive companies and, you know, if they didn't do that, they would have missed the boat. Someone else would have, uh, eaten that lunch. Uh, I don't know. Th- I'm thinking about the Technology Bros podcast on Twitter right now, TBBN, where they basically figured out on Twitter you can create this like live stream and you see it all day in your Twitter feed just like, "Hey, they're broadcasting," and it's a really cool distribution channel. So, so I think there's like a big call to arms here almost, of just, uh, the opportunities emerging and you basically need to pay attention. You can't opt out.

Brian Balfour 29:58 That's right. That's right, exactly.

Lenny Rachitsky 29:59 Okay, let's chat. Yeah. So let's

Lenny Rachitsky 30:01 talk ChatGPT.

Brian Balfour 30:02 So look, like, w- like, let's go through this cycle. We are right now, we're in that competitive environment. Like we said, like all those players we talked about, ChatGPT, Claude, Gemini, all these folks, they're, they are battling it out, right? Uh (laughs) and, and we've seen this with, you know, the talent wars especially over the past, you know, month, month or so. And, and so there's no clear winner yet, but there's consensus around, uh, the category.The second thing is then, okay, what's the moat? Has the moat been identified and who seems to have identified it the first or is furthest along? I think there's, uh, you know, my hypothesis, and I think there's a lot more consensus around this now than there might have even been three months ago, is that the moat is really about, is about context and memory. You know, these models, uh, you know, by themselves, if you compare them side by side, uh, you know, they, they kind of generate the same results. And so the, the actual difference maker is which one has more of your context and m- and, and, and because it's the context plus the model that produces the best output. And then, and then that kind of starts to accrue to this loop around memory. The more you use it, the more it's able to store memory around you, which kind of feeds more personalized context, which produces better outputs, right? And it ends up being an, uh, you know, another one of those flywheels, another one of those, another one of those loops. And so if you look at who's farthest on this, it definitely is, um, ChatGPT, right? Like they were kind of the first ones to memory. They've been investing a lot in these different types of, uh, data connectors, essentially context connectors, you know, e- you know, gathering, uh, all of this context. And, uh, and, and so you can really start to see it in the usage. But the second thing is, is in one of the pushbacks I've gotten on my prediction, has been, "Well, what about like Google and Gemini? Like, they have so much distribution through Chrome and like all of this other stuff," right? But, uh, it was, uh, Didi Dos, who's a, a VC at Menlo Ventures, um, actually, uh, published some good data on retention of all of these different ones. And, uh, I think the second, uh, reason I predict ChatGPT is like if you look at history once again, it was never the person who had the biggest distribution at the moment of time. It was the one that had the best retention and engagement. Google had the best retention and engagement over the others, Facebook had- it was smaller, but had way better retention and engagement over the others, right? So on and so forth. And so, um, the, the data that Didi published, uh, clearly showed that both the retention curves, uh, which I know you and I have both written about at, you know, at exhaustion, um, level off at significant portions higher than all the other platforms, as well as those retention curves have been shifting up dramatically over time. You can start to see, uh, the effects of, of memory and t- and they have the very elusive smile curve, right? The ones that you just like... And, uh, the only other times I've seen, you know, all of those dynamics, very few times in my career and they tend to be the folks like Slack and, and like all of the big winners. It's, there, it's just like so elusive.

Lenny Rachitsky 33:04 And the smile curve, just so, just so people haven't done it, that is essentially, retention goes up over time. It goes down a little bit and then-

Brian Balfour 33:10 Yeah.

Lenny Rachitsky 33:10 ... you come back to it and you use it more.

Brian Balfour 33:12 Yeah, that's right, and that's usually the result of, um, some type of network effect or, or something else. And, uh, and it, and it's, it's a early indicator that, uh, these fo- that, that that platform is on a trajectory to, to hit escape velocity. Um, the third piece is that, and they haven't really hidden these, uh, but there's all sorts of signals that they're about to launch a third party platform. Um, they've, uh, they've been hiring for a bunch of roles. I've seen multiple postings on like product manager, engineering roles, all that kind of stuff for, you know, quote unquote agent platform, a- and all those pieces. And so, um, it feels pretty inevitable that they will... one of these players will need to launch a third party platform in order to, you know, serve all the possible use cases o- you know, on these tools. There's gonna be some value exchange which is like, "Hey, for your agent to be effective, you probably need access to context and memory and distribution," right? So th- there'll be some value. She said, "Integrate to us and we'll give you those three things," right? Which is gonna drive more users and more usage and, and we're gonna go through the steps of, we're gonna go through the steps of the cycle. And you can already see this, right? Like, uh, you know, they're starting to form preferred partnerships, right? With, uh, some of the bigger players, which paves the way for, um, smaller third party players. It kind of gives, lends credibility to the platform. It's like, "Well, if HubSpot and XYZ are doing it, then I should probably do it too." It's, it, it's like that type of, that type of mentality. But that's why I think out of all of these platforms, ChatGPT has, uh, the, the best shot right now. And then, and then a bunch of folks are always like, "Well, well what about Claude? I really like Claude. I use Claude." Well, um, the problem with that is, like I think ChatGPT at this point has, f- s- like at least a 10X difference on MAU. So if you're a developer, right? And you're comparing those two platforms and you're saying, and you're looking at it and you're like, "Well, ChatGPT has 10X the number of users and better retention and engagement," it's like what's the, what's the logical choice of which one you're going to prioritize your scarce resources on, right? Um, and so, so th- those are just some of the reasons that my prediction is on ChatGPT and, um, in, in the blog post that I wrote about this, I actually then played my own devil's advocate and said, "Okay, here are some reasons why it might not be Chat, ChatGPT." But, uh, but I think we're in that part of the cycle. That's my prediction. I might be wrong in the prediction of ChatPT, ChatGPT, but I really think, uh, feel very confident we're gonna see this cycle play out again.

Lenny Rachitsky 35:45 Two follow up questions here. One is, what's your, what would be the backup if it's not ChatGPT? It sounds like it might be Gemini or Google.

Brian Balfour 35:51 My hypothesis of who's best positioned but is not executing on it right now would actually be Apple. Because-

Lenny Rachitsky 36:00 Whoa.

Brian Balfour 36:01 ... through the, the devices they basically can see everything. So they have the ultimate view into your context, right? They're, they're, they're sitting at that, they're sitting at that level. But I don't know what they're doing. (laughs)

Lenny Rachitsky 36:15 (laughs)

Brian Balfour 36:16 Like from an execution standpoint, maybe they're gonna surprise us with something crazy magical, uh, but oh, I haven't, we haven't seen any external signals around this. So, so that's probably...... just based on, on what real estate and where people live in the stack would own. And then I think right behind that I would probably put, I would probably put Google, uh, because of owning the context of things like email and dis- the distribution points of search and Chrome and Android and, and those types of, uh, pieces. But, um ... And a lot of people point to them, but, uh, my experience with all of their products is, uh, like going back to the retention engagement thing is, um, is that if we could take a look inside their metrics, I think what we would see is a bunch of flyby users in their miles. Like, they're kind of sprinkling the Gemini bucket everywhere. And it ... I'm a- I've, I've like literally clicked on it accidentally multiple times. (laughs) And so my guess is a huge portion of their miles is, is exactly that, uh, of like what, what's happening right now. And so, you know, look, they just, um, uh, they, they just acquired a very talented team from, uh, from Windsurf, and-

Lenny Rachitsky 37:30 Just the team, just the team.

Brian Balfour 37:31 Yeah.

Lenny Rachitsky 37:31 Part of the team.

Brian Balfour 37:32 We'll see. And, and things are, things are changing dramatically, you know, uh, on a, on a week-to-week basis. So we'll see if they're able to press those advantages in a very clear way. But I think the window is very small for them if, if Ch- if ChatGPT plays their cards right because they clearly have the escape velocity right now. And, uh, if they just keep pressing that advantage in, in the right way, I think it's going to be very hard for Google to, um, counter in, in the amount of time that's left.

Lenny Rachitsky 38:00 On the Claude piece, I'll just throw this nugget out. I had, uh, Mike Krieger on the podcast, head of product CPO at Anthropic, and asked him just, "You're, you're losing to ChatGPT. What do you, how do you approach, uh, the future of Anthro- of Claude?" And he very specifically said, "Yes, they've caught lightning in a bottle, and this is just going to win based on what I've seen-"

Brian Balfour 38:20 Yeah.

Lenny Rachitsky 38:20 "... at Instagram." And so we are specifically focusing on what is Anthropic and Claude incredibly good at, which is developer tools, coding, backend stuff. So they're actually leaning more and more into that. And if you've seen their revenue recently, they're making, I don't know, like approaching 10 billion a year or some crazy amount of money. So they're actually doing super well, just in a different use case.

Brian Balfour 38:40 Yeah, I'm glad you mentioned this 'cause this brings up something that, uh, we skipped, which is, uh, there are smaller platforms that, uh, have existed and will also emerge in this environment as well. And, and that's kind of like what you're alluding to, is, i- this tends to happen is, like, things end up, um, you know, uh, you know, growing into, into more niches. Like, e- even if you look at social, right? Uh, like LinkedIn emerged as a subset of, of the social world. But I'll, I'll, uh ... But even on these smaller platforms, these new distribution channels, they, they go through the same cycle. I'll, I'll give something, you know, really, uh, uh, a, a, a very opposite example of the ones that I gave. Like, lo- look at the platform Udemy, right? They, they are a platform for course creators, right? I don't know if most people know this, but when they started, their rev share to creators was something like 80% to creators. They started very high and ... Now, th- that brought on all the course creators, got their whole marketplace going, like, so on and so forth. I believe it was about a year ago they announced that they're essentially pushing that rev share down to something like somewhere between 15 and 20%.

Lenny Rachitsky 39:48 Wow.

Brian Balfour 39:48 They're somewhere like 25 and 30%, right? So I ... Another example, right? Of like, they closed down organic distribution in order to monetize, like, all that kind of stuff. And the same thing will happen in this AI world. I believe, you know, Cursor ... It's very clear, like Cursor's on the path to also probably create some type of agent platform, right? For developers. So that'll be like a smaller ecosystem to play in for, um, some, some products. Um, there's all sorts of ... Uh, it feels like everybody has the same strategy at this point, is everybody wants to launch an agent platform. I imagine some of these other horizontal productivity tools will do the same thing, maybe like a Notion or an Airtable or like a monday.com or s- or something like that. So there, there are s- ... There will be smaller platforms that will emerge, and they will follow the exact same cycle that I am, uh, that I'm also discussing. Um, but yeah, in terms of like the biggest kind of consumer one that, that's where I think ChatGPT has probably the, the most escape velocity, and, uh, and yeah, others will focus on different areas. And, and just to be clear, I love Claude. I actually u- use both Claude and, and ChatGPT-

Lenny Rachitsky 40:58 Same. Same.

Brian Balfour 40:58 ... all for, for different things. I, I have lots of love to go around for all these tools. I'm not ... My prediction has no, no bearing on which, which product I, I like the most right now.

Lenny Rachitsky 41:07 I also love Claude.

Brian Balfour 41:08 (laughs)

Lenny Rachitsky 41:08 So I think what ... So the key point here you're making is that there's almost, uh, a number of distribution channels emerging. Many of them will be niche. So I think of LinkedIn, if I wanna ... Like, LinkedIn for me is a very targeted audience for folks that listen to this podcast.

Brian Balfour 41:23 Yeah.

Lenny Rachitsky 41:24 So even though it's not, I don't know, Google or, or Facebook or whatever, uh, it's still incredibly valuable for the specific thing that I do.

Brian Balfour 41:30 For sure.

Lenny Rachitsky 41:31 So I think this is even more interesting that there's gonna be a number of distribution channels that emerge out of this whole AI wave. The other thing I'll note real quick, you t- mentioned this idea of everyone's building agents. I just had Brett Taylor on the podcast, who's building Sierra, and we actually ... He made me realize why everyone's building agents partly. One is because the, uh, the outcome-based pricing that you can, uh, charge with agents is incredible because one, you can actually attribute their impact on your business's ROI. You can actually see this is saving an agent, uh, $15 because it solved a case. And it's, um, it's attributable and it's autonomous. It's just doing it on its own. So with that, you can charge per outcome. You can say, "We'll charge you a dollar every time it solves an i- uh, an issue." So the monetization opportunity is huge and the margins go up like crazy.

Brian Balfour 42:18 Do you ... Can I just ask a question about that?

Lenny Rachitsky 42:21 Yeah.

Brian Balfour 42:22 Do you think that has longevity, um, in the sense that, uh, that makes sense in the current environment that we're sitting in right now because people are kind of comparing these outcomes relative to what it costs them today with pure humans. But once again, competition comes in at some point, and so that feels like that creates a-... pretty ripe opportunity to, uh, you know, undercut and come into... And then you have like whole... Ah, like, the disruption theory kind of playing out as well. And so, um, it obviously depends on, like, the infrastructure cost and computer costs to, to, to run these things. But I just wonder how much of that is, uh, temporary versus, uh, versus something that will be long term.

Lenny Rachitsky 43:07 So you're saying that dollar will come down to like $0.50, $0.25? Or are you saying someone's going to come with a whole new business model and disrupt that whole approach?

Brian Balfour 43:15 More the first. Yeah, like it's just competition erodes that away-

Lenny Rachitsky 43:18 Oh, okay.

Brian Balfour 43:18 ... essentially.

Lenny Rachitsky 43:19 Yeah.

Brian Balfour 43:19 Right, yeah.

Lenny Rachitsky 43:19 Yeah, yeah, that's a good point. So margins will be higher for a while and then they'll come down. Uh, assuming-

Brian Balfour 43:24 Yeah, un- unless-

Lenny Rachitsky 43:24 Yeah.

Brian Balfour 43:24 Unless there's, uh, something else that, um, creates a, like, durable pricing power, right, uh, in that case.

Lenny Rachitsky 43:32 I, I wonder if that... Yes, I wonder if what that is is-

Brian Balfour 43:34 So that's probably the second piece of this. Yeah, yeah, that's probably the second piece of that hypothesis, I, I feel like, yeah.

Lenny Rachitsky 43:38 Yeah, I think... I guess the opportunity there, the moat would be the data. Like, similar to how Cursor is collecting more feedback on what is... people want in their code suggestions. Maybe in theory, CRFN has more and more data over time and there's kind of this network effect where they build their data ?

Brian Balfour 43:53 Yeah, that's right, yeah. So, so let me revise that. Yeah, it's like, uh, I believe in that as long as it's paired with-

Lenny Rachitsky 43:59 Some moat.

Brian Balfour 43:59 ... uh, this, the second piece. Otherwise, it gets competed away, yeah.

Lenny Rachitsky 44:02 Mm-hmm. Good tangent. Okay, uh, one more question. What's the... What is your prediction on timeline for when things... when the opportunity appears, and what do you predict, as of the day we're recording, what do you predict will be the next couple things that open the ChatGPT, and let's just focus on that, releases, to start to open up this platform to get everyone in there?

Brian Balfour 44:26 Yeah, well, look, I, I'll first, uh, uh, give the disclaimer that I feel like any thoughts on timing in the AI market are very (laughs) ... have been very hard to pre- it there... it's always sh- shorter, that's where we should buy this.

Lenny Rachitsky 44:42 Mm-hmm.

Brian Balfour 44:42 It's always shorter than you think, uh, of like when something is gonna happen. Like that's what it's felt like from the seat that I've

Brian Balfour 44:47 been sitting in. Um, and, uh... But my guess is this: w- we're going to see the, the next major steps of this play out over the next six months. And so, I think we just saw one of the pieces drop around this, uh, which was, uh, the re- which was Ch- ChatGPT's recently launched Agent mode. And so it's kind of a general-purpose agent, you know, and, and I think that starts to introduce all of the users to, uh, to, to using agents and they're kind of figuring out and placing it in the different tiers and business models, a- all of those pieces. But, um, but it's likely that no general-purpose agent is going to fulfill all of the infinite use cases, um, successfully. And there's two reasons for this, right? Users struggle with horizontal tools, they can do everything and that's exactly why they struggle to adopt, and so they typically need more specific entry points. But also, uh, the more specific use case you get, sometimes you need specific UI, specific data, like other specific ingredients, you know, to properly fulfill, uh, that use case for, uh, you know... uh, for a given audience. And so I think their Agent mode was a step in this direction. What I would expect to see play out next is that

Brian Balfour 46:01 they will... they will either launch... they will announce the platform or what they're going to annou- with preferred partners or what they're going to announce first is, is basically a set of preferred partners, th- the guinea pigs. You know, an initial 10 to 20 folks that are like bringing agents, uh, to their platform, and, uh, and what that does is, it essentially... once again, it's a credibility card, right? You, you do special deals with, uh, some like right brand names to give, uh, the platform credibility, and it kind of creates this desire from everybody else to come on, you know, to the, to the platform. And then the step after that is starting to open up the platform. And this is where the real, uh, you know... where we'll really start to figure out what this game is going to look like because they basically have to define what the value exchange is. What are they giving you access to, right? Uh, and like what are the... what are they incentivizing you with t- to come on to the platform? So that's one version of it. Um, the other version of it is just like the replacement to search. Um, there will probably be... Uh, you can also see them starting to make mo- more moves here, which is, uh, like deeper attribution in some of the results, like, uh, those types of pieces. They're bringing in shopping, right? Like that's one of their recent announcements as well, kind of native into the UI.

Brian Balfour 47:18 Essentially, they will form new monetization mechanisms around that stuff as well. And that's actually going to be very important because, uh, for them to... Going back to the moat around memory and context, is that, uh, you know, they, they will want to incentivize as many people to their free, uh, tier as possible. But given the cost of AI, they have to cover it somehow, so they're going to need some monetization mechanism. So the more that they can cover that free usage with things that aren't subscriptions, uh, I think that probably also kind of feeds them as well. So I think those are some of the next steps on two different vectors, more with like a third party de- developer platform and more of the, you know, kind of content, whatever you want to call it, AEO, GEO, I don't know what acronym is th- every... we've all decided on yet, so let me know if we have. And, uh, and, and I think those will be... I think those will be the next... the, the next steps that we'll see. Now

Brian Balfour 48:14 that's what I think for ChatGPT. I think the thing that we should talk about is like, um, essentially what I would advise folks, especially startups, is you're placing bets. Y- at this part of the cycle, you're placing bets. We don't... The winner is 100% guaranteed, as I mentioned, and so you essentially, at some point, will need to make some decisions about, uh, you know, where t- where to place your bets. In the Facebook days, uh, all, all those other... all those other social networks, they also came out, you know, with their own platforms, right? And, uh, iOS had A- Android and some failed initiatives from Windows. I don't even remember what that platform was called, right? (laughs) Like... And you can look back and whoever placed their... You know, the iPhone was actually very... and iOS is a good one, which is if you had only aligned your bets to Android, you probably lost. If you somehow find- found a way to play on both ecosystems, you could be a winner. But if you only aligned to iOS, you could also be a winner, right? So like, that's just, right? Like, you, you had to have iOS as part of your betting strategy in order to win. So everybody right now, like, you're probably at this cycle and, and trying to figure out, well, y'all need to ... Everybody will need to figure out where are they gonna place their chips, how, how are they gonna bet. And depending on how you bet really depends on, um, what your current position is in the marketplace. You know, if you're a late stage startup, let's start with that, um, or like a late stage company, you can afford the luxury to place multiple bets and kind of spread your chips and, and kind of wait it out a little bit to see who the winner is and then really throw your muscle, you know, b- b- behind that winner. You, you have that luxury a little bit. And of course, but the risk of that, the, the, the risk of that is that the, sometimes the incumbents wait too long to, to make that decision, and, and that's like kind of the key, key question they will need to answer. The key question for startups is totally different. You don't have the luxury to spread your chips. Like, you have to go all in. You have to choose one and go all in. Um, you, you have scarce resources, scarce attention, um, from the market, and so it's a totally different ballgame. Higher risk, higher reward, for sure. Uh, and, uh, and, and, and that's part of the betting strategy, uh, for startups. And so, uh, that's kind of what y- that's kind of what you have to do, is you kind of have to figure out your betting strategy, and then, you know, we, we can talk a little bit about how you might evaluate and pick the right course for you. But, uh, but that's where we're all at right now, is we're, we're kind of, we're, we just entered the casino, uh, we just cashed, put some cash in for some chips, and now we, now we've gotta figure out w- you know, what tables and, and where to, to place those chips.

Lenny Rachitsky 50:56 I love this analogy. Okay, so just to be crystal clear about what listeners should do, what founders should do, what product teams should do, the advice here essentially is, uh, integrate with, and ChatGPT, maybe Gemini, maybe if Apple has something, is like actually integrate with, with what they launch. So could be, uh, a login thing, could be a search thing, could be a connect and suck up your memory and context. The advice here is, is you need to do this, 'cause this is potentially the way that most companies will start to grow, and your competitors may overtake you.

Brian Balfour 51:30 Yeah. Yeah, if we had to, like, really simplify it, it's essentially play the game. Don't opt out of the game. Don't, don't trick yourself into thinking that you can't play the game. That's number one. And then number two, no matter who you bet on, just make it a focus bet. Uh, 'ca- because the, the, all, like, if you look back, all the failures are the ones that tried to, you know, play multiple games at, at once with scarce resources, and that just tends to never work if you're, you're an early stage startup. So, those two things. Play the game, put a focus bet.

Lenny Rachitsky 52:03 This episode is brought to you by Miro. Every day, new headlines are scaring us about all the ways that AI is coming for our jobs, creating a lot of anxiety and fear. But a recent survey from Miro tells a different story. 76% of people believe that AI can benefit their role. But over 50% of people struggle to know when to use it. Enter Miro's Innovation Workspace, an intelligent platform that brings people and AI together in a shared space to get great work done. Miro has been empowering teams to transform bold ideas into the next big thing for over a decade. Today, they're at the forefront of bringing products to market even faster by unleashing the combined power of AI and human potential. Guests of this podcast often share Miro templates. I use it all the time to brainstorm ideas with my team. Teams especially can work with Miro AI to turn unstructured data like sticky notes or screenshots into usable diagrams, product briefs, data tables, and prototypes in minutes. You don't have to be an AI master or to toggle yet another tool. The work you're already doing in Miro's canvas is the prompt. Help your teams get great work done with Miro. Check it out at miro.com/lenny. That's M-I-R-o.com/lenny. So, s- you were head of growth at, at HubSpot for a long time, so, and you gave that as an example, HubSpot, why would they integrate, why would they give away all their data so that ChatGB can suck it up and you never have to go to HubSpot, you're just working through their agent? You're, would you at HubSpot be like, "Yes, we gotta do this. This is the game we gotta play"?

Brian Balfour 53:31 Yeah, 100%. And that's exactly what I think you see them doing, and I, look, to be very clear, I have not talked to anybody at HubSpot about this. I have not talked to Dharmesh about this. But, um, Dharmesh, I think has also, like, published, publicly po- published about this w- uh, but that the right thing to do is essentially, e- even though you don't, you, you understand how the cycle plays out, and you don't necessarily under what y- understand what your exit strategy is, uh, once you get out, it's better to be early, know that you need to figure out an exit strategy and figure out that exit strategy along the way, versus waiting and then being super late and, and then, and then know what the exit strategy is. And, and I think that's essentially, um, what y- that's exa- that's exactly what you see them, see them doing. They're trying to be as early to this stuff as, as possible, and I think it's a, I think it's a pretty smart play, even though we might not necessarily see, like, what, what the exit strategy is out of this cycle for them.

Lenny Rachitsky 54:34 So, going back to that amazing quote that you shared at the beginning of the conversation by, I think it was Alex Ram Paul?

Brian Balfour 54:40 Yeah.

Lenny Rachitsky 54:40 Of that startups win by finding a distribution channel before the incumbent copies them, and what you're saying here is this is the opportunity for startups to disrupt an incumbent. This is the opportunity for someone to disrupt Salesforce, I don't know, ServiceNow, all these guys that have been around for a long time.

Brian Balfour 54:59 Yeah. It's gonna be one of the major ones. Now, look, you've already see players that are, have been able to hit this escape velocity.... uh, you know, the, the cursors and stuff of the world. Um, and, and so, uh, there... once again, there's multiple ways to hit that escape velocity, but this is gonna be one of... this is one of the major ways to do it, is to basically hitch yourself to a new platform. Look, you did it yourself actually. You hitched yourself to Substack super early.

Lenny Rachitsky 55:26 I was gonna-

Brian Balfour 55:26 You took a focused bet.

Lenny Rachitsky 55:27 ... I was gonna say that.

Brian Balfour 55:27 Uh, yeah, yeah. I was like... I don't know why that just hit me, but you took a focused bet and you benefited from it in a disproportional way than those that kind of came later. Uh, and I think that's actually a great meta example here (laughs) as, as I sit here and think about this.

Lenny Rachitsky 55:44 Yeah. That's actually the way I thought about when I was moving to Substack. Just like, I feel like there's this wave rising and I want to ride this wave even if it... maybe it's not the best place or may... you know, they take a cut, all that stuff.

Brian Balfour 55:53 Yeah.

Lenny Rachitsky 55:54 Uh, but it worked out really well.

Brian Balfour 55:55 That's right.

Lenny Rachitsky 55:56 Uh, and, and-

Brian Balfour 55:56 I think it worked out very well (laughs) .

Lenny Rachitsky 55:58 It worked out really well.

Brian Balfour 55:58 (laughs)

Lenny Rachitsky 55:59 Uh, and to be honest, it, it felt like it was too late, the- when I started six years ago.

Brian Balfour 56:03 When you- when you entered, it felt too late?

Lenny Rachitsky 56:05 Yes. Yes.

Brian Balfour 56:06 Oh, say more about that.

Lenny Rachitsky 56:07 Like, it's... it always feels too late, I think, to people that joined. Mm-hmm. Like Silicon Val... or sorry, Marc Andreessen has this famous quote. He's like, "I came to Silicon Valley in the '80s. I thought it was over. It was too late. I missed all the opportunities."

Brian Balfour 56:19 That's fair.

Lenny Rachitsky 56:19 Um, yeah. So yeah, there were just a lot of newsletters that were doing really well, a million subscribers. I'm like, "Yeah."

Brian Balfour 56:24 And what do you say to people now who want to join Substack?

Lenny Rachitsky 56:26 That's, that's... Learn from this example. Uh, a lot of times when people think that it's too late, it's, it's definitely not too late and it's always only just getting started. Especially if you're like on Twitter all day listening to podcasts like this where we're surrounded with this bubble of everyone talking about something, when in reality, like 1% of people know anything about what you're hearing about every day.

Brian Balfour 56:45 Yeah. Uh, yeah, that's so interesting.

Lenny Rachitsky 56:47 Okay, so coming back to the advice. Say someone is sitting there and talking to their manager like, "Hmm, Ryan just shared all this mind-blowing advice. We gotta pick our battles. We gotta pick our platform." What would your advice be for them to decide where to, where to place their bets?

Brian Balfour 57:05 Yeah. So I think this is a great question because once again, you know, put my personal prediction aside for a second and, and I would encourage everybody to think about it from first principles, from their a- like who their audience is, what their product is, what stage of company they're at, like all those, all those... their current strengths and weaknesses. You gotta take all this into account. But if I had to boil it down to a few criteria, the main things I would think about is when you're looking at new distribution channels and new platforms, um, to choose on, one, kind of going back to what we said before, is the better signal is retention and depth of engagement of the users on this platform than it is like pure kind of user level, like MAU or s- some, some other, like number of signups or, you know, one o- one of those vanity metrics. So look at that, number one. Number two is, there's some element of like user quality and monetizi- and ability to monetize the users on this platform. I think the starkest example here would be, you know, iOS and Android. It's like even though even today it's something like Android has 70 s- 70 some percent of devices but only 30% of the market share by dollars and it's the exact flip, uh, for iOS. So it kind of goes back to what we were talking about earlier, which was, if you bet on Android only, you probably lost, but if you bet on iOS only, even though smaller user base, you could... you, you were still able to parlay that into a win, um, later. Um, the third thing to look at is, as these platforms emerge, just analyze what the value exchange is, right? So what are they giving you to incentivize you to develop on their platform. And, uh, and like all these platforms are a bit of a... it's a bit of a game of whoever understands the rules and how to arbitrage the rules the best, right, tend to be the ones with the e- w- with the edge, uh, and, and figure that out.

Lenny Rachitsky 58:58 Yeah.

Brian Balfour 58:58 And then, and then finally fourth on my criteria would be pure scale, right? Obviously, even if you have those other three but there's, you know, a 200X difference in s- in scale and, and momentum, then like y- o- obviously you probably have to choose, um, the bi- the bigger platform. Um, but last but not least is, as you go through these criteria, this is, this e- these are how you think through entering the game, you know? And once you enter the game, then you immediately need to move t- to starting to think about how do you exit the game. Knowing once again that, that, that last step is going to come at some point in the future, that there's going to be some closure, uh, for monetization, then that's where you have to start thinking through your strategy to, to exit. And that comes down to things like, okay, well, how are you going to own a important part of the user experience or workflow. Or, you know, how are you going to accumulate specialized data in context that the pl- major platforms don't have? Um, or how do you create, you know, different types of like micro network effects? Uh, like all, all of these types of things. So just once again, the- there is the entrance criteria, but once you figure that out and you feel like you're in the game, you immedi- immediately need to move towards, okay, you know, what, what, what's my exit plan here knowing this is all coming?

Lenny Rachitsky 1:00:22 It's interesting that another way to think about this model you've described is building... the strategy of building on top of LLMs and becoming a GPT wrapper. Because essentially, this tech allows you to, say, create a cursor that is incredible, and then you could argue, oh, you're just gonna be this wrapper and why would anyone... like they're getting all the money here, everyone can copy you. Y- like how- what's your defensibility long term? And the answer is, what is the moat you will build over time sitting on top of this thing that will make you more and more valuable long term and not have to rely on this thing? So it feels like you could use the same framework for building a GPT wrapper business.

Brian Balfour 1:01:01 Yeah.

Lenny Rachitsky 1:01:01 To use that euphemism.

Brian Balfour 1:01:02 Yeah.

Lenny Rachitsky 1:01:03 Um, so say someone is sitting there today. Is there anything they can do to start making a bet? Is it simply creating an MCP that allows...... LMs to suck in your data. Is that the one thing you could do today? Is there anything else that's available today to start, uh, using these platforms or is just a little too early and they haven't released the good stuff yet?

Brian Balfour 1:01:25 It might be just like a tad too early, like we're like right on that edge but, um, you know, some of the questions I'm asking myself is I, I, like I'm kind of going through all of these, these players and where our customers and target audience live, and I'm asking myself, you know, the question, okay, if this player launched, you know, some type of platform, like how would we evaluate it? You know, uh, so on and so forth. It's hard to... Uh, you can also try to cozy up, uh, to these folks. I would place a large portion of my net worth right now, that if we could sit in the OpenAI offices, like at that front desk, that they are having, you know, meetings with (laughs) potential preferred developers, like talking about this. We could probably sit there a- and, and, and log it. So I do, do think some people are going to be in a place to develop, uh, preferred relationships and, and make a note. And if you're, you're in that spot then you, you should definitely play that card. Um, a lot of early stage starts will be in that card place. Um, other than that, I would say that w- we still need... Like, once they launch these platforms it's like we, you can't do much else until you really kind of know what the, the value exchange is and what they're gonna expose, uh, for you. But also, just be prepared to turn your strategy on a dime and go all in. I think that's probably one of the hardest parts of this is that, uh, these things emerge and you have to capitalize extremely quickly, and a lot of times it's just, it's hard for leaders to do that 'cause they don't want to create a feeling of whiplash, into the unknown, we've got all these projects in pla- like all, you, you know all the things. (laughs) So I think that's, uh, I think that's probably the, the last part of what, what we can be doing right now versus just, you know, kind of staying on top of everything as, as it emerges.

Lenny Rachitsky 1:03:14 As you were talking, this reminded me, I recently noticed that ChatGPT is driving me to my newsletter more traffic than Twitter, and I feel like that recently shifted. Uh, I didn't even know this was a thing until I just started looking through my referrals, I'm like, "ChatGPT? What the hell is going on here?"

Brian Balfour 1:03:31 Yeah.

Lenny Rachitsky 1:03:31 And I think it's like a different version of what you're talking about but essentially it's, like in theory I can block ChatGPT from, I don't know, I don't even know if I can from being read all my stuff.

Brian Balfour 1:03:39 You can in F- you can in Substack now, yeah. I just saw that setting in there, yeah.

Lenny Rachitsky 1:03:42 Okay, okay. Okay. Oh, interesting.

Brian Balfour 1:03:44 Yeah.

Lenny Rachitsky 1:03:44 But that's the similar kind of decision is like is it better for me for it to be recommending my stuff and telling people, "Hey, go check this thing out," or is it better to block it off? And I think, uh, per your point, I, and this is how, why I felt, I'm like take it all just like it's good. It'll... Yeah. It's r- but...

Brian Balfour 1:04:00 (laughs)

Lenny Rachitsky 1:04:00 In that, it's better that it's like from Lenny's newsletter than something else. So someone else will come in and eat that market share, in theory.

Brian Balfour 1:04:07 Yeah, that's right. Yeah, like if you, if you don't do it's a, uh, some- you know, somebody else is. Um, and I think, and I think that's also kind of what all the major media publishers are really contending with right now.

Lenny Rachitsky 1:04:18 I guess I need a licensing deal in your checks though. Anyway, okay.

Brian Balfour 1:04:21 Yeah.

Lenny Rachitsky 1:04:21 I want, I want to go on a totally different tangent. We weren't planning to talk about this. I know that, uh, I said this was gonna be fully focused on this one topic but there's something you mentioned to me before we started recording that I think will be really interesting to a lot of people. So

Lenny Rachitsky 1:04:34 you guys at Reforge are now building actual SaaS products that people can buy. It's not just courses. I don't know if people know that but let's make sure people understand this. There's actually products for product teams so maybe just explain that briefly but the thing that I think is really interesting here is you work with a lot of companies now, selling them AI tools and you have noticed a very big difference between the companies that are really good at adopting AI tools and seeing gains from them from those that don't. Talk about just what you see there and because this is, in theory, gonna be really helpful to companies that are struggling with adopting AI tools and seeing gains.

Brian Balfour 1:05:09 Yeah, just to quickly explain that tran- transition so it, uh, it makes sense for people which is, you know, I started Reforge just with the interest that there was all these incredible leaders out there growing, you know, on the front lines of some of the fastest growing companies and they have all this amazing knowledge and I wanted to encode it in useful and practful, practive, you know, practical ways for others, right? And that took the form of courses, uh, and content and product, all that kind of stuff at the beginning and along the way everybody kept asking us to essentially build the tools to implement what we taught. Um, because, you know, with anything is like you can learn as much as you want, you can listen to my podcasts, your podcasts, Lenny, like whatever as much as you want but if you don't actually put it into action and implement it then it's not really going to create value, right? And so people kept asking us to, to really close that gap and we said no for the longest time and then about a couple years ago when AI really started to inflect it really created this moment that, oh wow, now we... There's this opportunity not just to encode this knowledge into content but also into the products, the software, the tools that we use ourselves. And so we started to take a really big bet on that and, uh, and started to develop this, this new, um, platform for AI native product teams. The first product we launched was called, is called, uh, Reforge Insights which acts like your AI product researcher. Kind of aggregates all the feedback from all the sources, uses AI to analyze it, helps you explore it, um, but also, uh, will start to identify like what are the gaps, the things that you don't have in your feedback today and auto generate the research to go gather all those n- new insights so complete the full cycle. We're gonna launch two other major products as part of this platform before the end of the year but, uh, we'll, we'll save that for some, some future episode. So that's kind of been our, our journey and so we've seen inside companies that are going through this transformation from two perspectives. One is obviously selling in that tools but the other perspective is for ten years companies have been coming, uh, to us to help them drive some sort of transformation with our learning product. Uh, you know, most people, most companies are not coming to us to,... uh, just like throw a bunch of courses in front of ... They're, they're trying to solve some big business problem, some transformation. Now that used to be things like, we've got to figure out this growth thing, right? Or I'm going from sales-led to product-led, or, um, you know, I'm turning, uh, I have, uh, more project managers and I need to transition to product managers, right? Uh, like something like that. Like, or, uh, there's some business problem. They're going through some transformation and they, they saw us as part of that transformation, and we got to partake in quite a few of those types of transformation. Now, of course, the transformation that everybody's going through is, "Okay, how do I become more AI native? How, how do I adopt this stuff?" And, and so we've seen a pretty wide spectrum and from both perspectives of how companies are approaching this. And I'm sure everybody's seen the, like, AI ... oh, I call it, we've been calling them, like, the AI manifesto memos from CEOs out there that proclaim, "We are now AI native." (laughs) You know, uh, in some grandiose, you know, way, right? Uh, and, uh, but there's behind the scenes, there's actually some incredibly stark differences in, uh, the actual teeth of what backs up those memos and backs up those executive decrees that, you know, we should all be, uh, you know, AI.

Brian Balfour 1:08:41 And, and so just to kind of point out, um, a few of them, which is, um, one is that, uh, they're ... I think the most impactful thing, um, that you can do is form really hard constraints. So a lot... what I... what you see... there's, there's other parts is like, okay, you wanna communicate this, you wanna establish an owner of who's going to drive this, uh, you wanna build in incentives and rewards. And you see this all playing out in things like, um, you know, building it into your, uh, you know, your career ladders. Or some people are starting to introduce this as questions into, uh, their performance reviews. Like, uh, you know, all those types of pieces. And, but the thing that is actually moving the needle are the companies that are defining incredibly hard constraints. So one company that we worked with developed this constraint that they benchmarked against other companies of their revenue size and the team sizes, uh, for their stages, and they set a benchmark that we will be one-fifth. Each of our functions will be one-fifth the size. And, and what that did is it created a constraint that you couldn't hire above that level, and it forced people to essentially find ways to adopt AI and, and do things to, to replace that. So that was one. Um, you've seen these other ones. I can't remember from what company. This might have been Shopify or another who was like, "You are not allowed new headcount until you prove to us that you are not able to accomplish this, uh, with AI." That's like another hard constraint. And but you also see these other constraints on a smaller level, which is, you know, an executive saying, "I will not do a product review, uh, or review a PRD unless it comes with three prototypes." You know something like that. And so that's, that's the hardest one. Uh, those are the biggest constraints. But I think the biggest change that I'm seeing is a- and the things that separates the, the top few percent making this change and, and everybody else is, uh, is essentially making the hardest decisions. And that hardest decision is going to come down to exiting people. So in every transformation, what we see is essentially three groups of folks. You see your, uh, we call them the catalysts, the people kind of leading the charge, the people who are experimenting, you know, doing this on their own, um, time, like all that kind of stuff. You then have, uh, your, what we call your converts. Um, these are folks that will make the transformation, they will adapt, but they need structure. They need permission. They need a clear outline. They need a clear plan, right? And I don't say this in a negative way. It's just that, you know, that's how some people operate, right? And, and so you, um ... And so that's where things like, uh, all the things that we were talking about before which was like, the decree, the permission, the clear budgets, the rewards, like all of those types of things. Um, but then inevitably you have a certain percentage that are anchors, right? And they're dragging their feet. They're kind of, uh, you know, silently creating friction in the background, uh, and like all those pieces. And there's a big difference in how I think companies are treating, uh, and, and thinking about their strategy for those folks. One group is kind of like, "Oh, we're gonna work with them very passively." Others, others have set a hard deadline. They're like either make the ... They're gonna either make the transformation by X date or we're going to exit folks. And a lot of people look at this as being really harsh. I think a lot of people would think that, especially individuals. But let me kind of explain it from a, a s- more of like a CEO perspective. Um, a lot of these companies are seeing this AI transformation. The, the ones that are taking it more seriously, as this isn't adopting new tools. This isn't a light change. This is a fundamental culture change of how we operate as a company, right? And you can't have 20, 30%, whatever meaningful number of it is, of your company trying to operate in a completely different way, in a, in a completely different culture. Cultures thrive on density, right? And that's why they're sometimes the best ones feel like cults, you know? Um, and, and so as a result from that perspective, it's like, hey, like, we ... For us to be successful, for this to be the best thing for all employees, we all need to be operating around the same cultural principles and stuff. And if that's not you anymore, then we're defining a plan to exit it. But I would say that less than 10% of companies we see are taking this hard stance. But I would say they are probably the ones that are farthest along, getting the most adoption, and are seeing the most results, uh, of the ones that are-... taking those hard stances. So, there's a bunch of other stuff I could talk about, but that's kind of the high level of, kind of what we've, w- what we've seen across a bunch of different companies.

Lenny Rachitsky 1:13:37 That is incredibly interesting. I'm glad we went there. There's a, I have a newsletter post coming out soon, probably before this episode, that touches on some, a lot of advice along these lines. Um, I, I, I'm excited for you guys to keep sh- seeing these, uh, insights into companies and sharing more of those. 'Cause this is, I think, what a lot of people are looking for. Just like, things aren't quite clicking at our company. We keep hearing everyone is getting so much more productive. All these companies are, uh-

Brian Balfour 1:14:02 Yeah.

Lenny Rachitsky 1:14:03 ... running more efficiently, and, and it's not working here. And so, I think that's the kind of advice a lot of people are looking for.

Brian Balfour 1:14:08 Yeah.

Lenny Rachitsky 1:14:08 So, thank you for sharing all that. Ryan, is there anything else that you wanted to touch on? Anything else you wanted to leave listeners with before we get our very exciting lightning round?

Brian Balfour 1:14:15 Well actually, just a couple more points on this topic-

Lenny Rachitsky 1:14:17 Oh, yeah.

Brian Balfour 1:14:17 ... I think we should, we should go. Is like, um, uh, is, there's probably two more

Brian Balfour 1:14:23 things I would say about this. One is that, um, so if you're a CEO listening to this, I would say that most CEOs or most executives are incredibly disconnected from the actual AI adoption taking place in, in, in inside their companies. I think a lot, a lot of executives who have done these dec- decrees and all that kind of stuff think it's kind of happening naturally. But we talked to both groups. We talked to tons of end users and we talked to tons of executives. The story we hear from the end users, the PMs, the eng., all that kind of stuff, that we talked to kind of using all this stuff, one of the main questions we