Skip to content

By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.

Product SchoolProduct School

Bolt CEO on How PMs Turn AI Prototypes Into Production Code Engineers Trust | Eric Simons | E311

One month before Bolt exploded, Eric Simons was preparing to wind the company down.

Eric SimonsguestCarlos González de Villaumbrosiahost
Sep 9, 202631mWatch on YouTube ↗

Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.

EVERY SPOKEN WORD

  1. 0:00 – 0:44

    Trailer

    1. ES

      People are not losing their jobs because AI, at least certainly to the degree that some AI leaders have said it was gonna happen. What's actually going on though is it's like

    2. CV

      I was driving around and I saw a billboard sold by your company that said [beep]

    3. ES

      Yeah, yeah. Team was like 10 or 12 people at that time, and we, overnight we just woke up with tens of thousands or hundreds of thousands of paying customers.

    4. CV

      Eric Simons, CEO of Bolt.

    5. CV

      I, I saw a post you made a few weeks ago that says, "Best advice is don't die." I think that is so powerful.

    6. ES

      We launched in October of '24. You know, in the first month we went from zero to 5 million of ARR.

    7. CV

      Back in the day, the whole winning strategy was you have to be hyper, hyper specific. Analytics, prototyping, road mapping, and now it's about a platform.

    8. ES

      There's only two ways to make money in the software business, uh-

  2. 0:44 – 1:22

    Intro: Carlos welcomes Eric Simons, CEO of StackBlitz (Bolt)

    1. CV

      Hey, this is Carlos, CEO at Product School and your host on the Product Podcast. My guest today is Eric Simons, CEO of StackBlitz, the company behind Bolt, and I'm a user of the product. He skipped college, spent seven years getting to half a million dollars in ARR, and had a board meeting scheduled to wind the company down. One month before that meeting, they launched Bolt and went from half a million dollars to five and a half million dollars in 30 days with a team of 12. Things we'll cover: the pivot that happened one month before the shutdown meeting, why the cloud ID market turned out to be a mirage, his Yahoo and AOL warning for everyone building everything, why he signed the open weights letter, the billboard apologizing for the SaaS apocalypse.

  3. 1:22 – 1:37

    Welcome, and being a user of the product

    1. CV

      Let's get into it.

    2. CV

      Welcome to the Product Podcast, Eric.

    3. ES

      Thank you for having me. I'm excited to be here.

    4. CV

      And excited to have you because I'm a user of your product and it's always great to kind of put a, a face on the, on the name and, like, learn more about the story behind the scenes.

    5. ES

      Heck yeah. Awesome. Well, yeah,

  4. 1:37 – 2:07

    The Iron Man hat and going all-in on sports

    1. ES

      excited to dig in.

    2. CV

      So last time we spoke you were wearing an Iron Man hat. I know that you are big into sports, so maybe we can start there.

    3. ES

      Yeah, I have my Iron Man hat over here. I, I can put it on if you want, you know. Yeah, yeah. What, so yeah, what do you wanna know about, about, about sporting?

    4. CV

      So [laughs] I'm obsessed with sports too, but I think you kind of take it one step up. There was a story I think that connects also with your upbringing. I know that you are a high school dropout and, like, you've been always looking for hard path in a way. So maybe tell us more about what is that thing that is driving you so much to

  5. 2:07 – 4:38

    First principles, and finishing high school before dropping out

    1. CV

      not give up.

    2. ES

      Yeah, I think for me, uh... Yeah, so my, my parents would've shot me if I had dropped out of high school. So I, I, I did, I finished high school, but then I didn't, uh, instead of going to college I started doing startups. And yeah, I think, I think to me, I don't know, it's a good question. I, I think that I guess I tend to really look at how I wanna approach life, I guess, very much like a first principles sort of way and that's like a very, kind of an overused term at this point. But I mean, for where I grew up, um, the server Chicago, kind of the, the thing is, is certainly back in the late 2000s and, you know, it's like you graduate high school, you go to college, you get a job. I mean, that, that's like, that was like the path. Uh, there was no other kids that I grew up with that did not do the college thing or whatever. But, uh, I, I just kind looked, you know, at, at the time, you know, uh, my co-founder and I, he and I actually grew up down the street from each other. I think we met when we were like 13 and we learned how to code together. And so by the time we came to graduate I was like, [laughs] "You know, we're gonna have to pay for college out of pocket." You know, it's, it was gonna cost, you know, for, uh, University of Illinois at, uh, Champaign Urbana is like $30,000 a year in state. And we were like, "This doesn't make a lot of sense." I mean, we're making, at that time I think it was like maybe we're making 50 bucks an hour at that time at 17 or 18 writing software. And so it, to me it just seemed kind of obvious like w- what am I gonna go to college to go and learn? Like I'm writing software, I certainly can learn, but I can read textbooks, right? [laughs] And that's free. It's kinda like the Good Will Hunting thing, right? Um, [laughs] but um, not, not that I'm anywhere close to the IQ of that character that Matt Damon played but, um, but yeah, that, to, to, to me it was just, it was just going and doing what s- what seemed kind of obvious and then that's now kind of priced in. Like it's, you know, the whole college thing is, um, is the default path. Uh, m- many people, uh, are questioning that, um, which I think is a good thing. But yeah, and then, you know, especially as you, if you're gonna do business, you know, like starting startups it is, you're, you're not gonna have a good time with it. You'll quit pretty early on if you don't develop a lot of grit and resilience because the, the, that's kind of the entire game. This is an endurance sport and it is about pushing through pain, right? And, and, and that's like, um, so a couple years ago I got into and I started doing Ironmans and that sort of thing and it's the, it's very analogous where you, you have to be going for very long stretches and, um, it's not even about race day it's about all of the training that goes into it. I mean, you're, you're spending six hours a day each day on every weekend for six months leading up to that thing, you know? So, um, and then such as, again, such as startups where it's not about one big launch or whatever it's, it's, it's about every day you're putting in the work and, you know, embracing pain and, and, and pushing through it. So that's, that to me is kind of how I view

  6. 4:38 – 5:07

    The link between endurance sport and building a company

    1. ES

      these things.

    2. CV

      I love that connection that you made between, you know, like sports and, and business. I kind of feel the same way. Um, I, I saw a, a post you made a few weeks ago that says, "The best advice is don't die." And I, I think that is so powerful because sometimes they think the, the pressure is on like growing fast at all cost or, or just winning the short term. But in reality like staying in the game, finding ways to not die is way more powerful and compounds much more than just trying to

  7. 5:07 – 5:54

    Cutting through the short-term noise in the space

    1. CV

      win the sprint.

    2. ES

      Yeah, totally and, and over the past year and a half there's been a lot of, a lot of short term in the space we're in. There's a lot of short term strategies being done and, and e- every time this happens in business there, there's kind of these moments where you have these manias where you, you can, you can do very shortsighted things and maybe walk away financially. Uh, the problem though that, uh, i- from my view it's, it's, it's what you're talking about. It's like if you're gonna build a business why do this if you're not trying to build something that's actually durable for the long term, right? Like you, you gotta, you gotta do it like the right way, you know? Because o- other- otherwise it's a pretty high risk endeavor for you and certainly for anyone else that's working for you, right? So yeah, so that to me is that, you know, I think business and life it's, it's, you know, kind of, um, the, the Naval quote of like, you know, play long term games with long term people, right? Like that, I think that's just, you know, it's a good quote and I

  8. 5:54 – 6:17

    From StackBlitz to Bolt: the origin and the name

    1. ES

      believe in it.

    2. CV

      Let's talk about your, your product and your business. I mean, even your current company it's called Bolt, but it wasn't the original name, right? StackBlitz was the original name and you actually started the company before AI was mainstream. So-

    3. ES

      Yep

    4. CV

      ... tell me a little bit more about that major pivot that you had to do in 2014 to kind of put the posi- Take the AI tailwinds and make sure that you are in the position that you are today with a scale of, like, $700 million in valuation.

  9. 6:17 – 7:49

    Seven years of bets before Bolt worked

    1. ES

      Yeah, I mean, so yeah, we've been around. So we started, uh, StackBlitz, which is the underlying company, um, behind Bolt. We had started in 2017, and the insight we had there was, um, the browsers had become very powerful, [lip smacks] and we realized that it seemed like it was technologically possible to bring, like, full-stack web development to the web. So in the same way that Figma had brought design to the web for the first time, the idea that you could come just in a browser, you don't have to download anything, you could actually be building real software, like, real full-stack applications from a browser tab, that would-- it was possible for the first time. And we wrote this really cool technology. We basically wrote an operating system that runs in WebAssembly in your browser. Very nerdy stuff, but, like, um, no one had ever done it before. And it took us, uh, I think, like, three, four years to build that technology. We launched it in 2021. Um, this was, like, in the cloud IDE market, and, uh, at that time, you know, in late 2010s, early 2020s, th- there's all this hype around cloud IDEs. But it ended up being a mirage of a market where there was a lot of hype, but people weren't willing to spend money. Like, developers were very happy with their local environments. And so fast-forward to 2024, uh, us and all the other cloud IDE companies were on the rocks. Um, we had, you know, had to figure out what we were gonna do, and, um, you know, for us, we had, we had set the target of we gotta inflect revenue. We, we, we gotta prove that this is a venture scale business here, otherwise we're gonna start winding this thing down by the end of the year. And, and so through 2024, we tried out a, you know, a whole bunch of different ideas, you know, of different kind of product shapes that we thought, um, you know, would be, uh,

  10. 7:49 – 8:42

    Launch: 0 to $5M ARR in the first month

    1. ES

      interesting and, you know, be-- customers would, would like and et cetera, and none of them worked. And Bolt was actually the, the last product idea that we tried, and we launched in October of '24. Um, our board meeting was one month later, where we were gonna start spinning down the company. And, uh, you know, in the first month, we went from zero to 5 million of ARR. You know, uh, th- b- before that, we had spent seven years getting to 500K, 0.5 million [chuckles] of ARR, and then in 30 days, we went from, you know, 0.5 to 5.5 million of ARR, right? And then the month after, we went from, you know, 5.5 to 20.5 or something. You know, so it was just, like, crazy, right? And, uh, y- our team was, like, 10 or 12 people at that time, and we-- overnight, we just woke up with, you know, tens of thousands or hundreds of thousands of paying customers, so it was-- Growing into that scale was, was insane. But that's, that's kind of the story of how, you know, Bolt came to be.

  11. 8:42 – 9:46

    What real product-market fit actually feels like

    1. CV

      Speaking of product market fit, you know, it's hard to pick an official definition, but, like, what you just said, when the market is pulling, that is ultimately the, the best validation.

    2. ES

      Absolutely. Absolutely. Yeah. It's-- I've worked on a lot of things in my career, and I've never, yeah, I've, I've never seen anything like what happened there. A- at that time, you know, there's a, there's a number of companies now that have had these zero to, you know, X explosive revenue ramps, but we were the first thing that hit the market that had ever seen anything like this. And, uh, so it, it's, it's, in the AI era, there's, there's now been more of these, but it was, to, to, to be the first one was an interesting experience because we were going and talking to people, uh, e- because, like, normally you have, you know, to, to the playbook to grow, like, the team of a company to w- that's at 20 million of ARR, usually you have minimum two years to, like, [chuckles] prepare for that. And we, it, it, it, we, that happened in two months for us. And so it, it, it was just, uh, th- there were no playbooks. And so, you know, we ended up, you know, having to create new playbooks on how to, how to deal with this sort of scale this quickly. So it's, it was a fun and, and very challenging experience, actually. [chuckles]

  12. 9:46 – 10:34

    The $700M valuation and the state of the business

    1. CV

      So, so moving forward, you raised over $100 million, valuation of, of around $700 million, right? So what is, uh, the current state of the, of your business?

    2. ES

      Yeah, I mean, I, I think for us, like, we're very focused on... You know, like, when we first launched, we had lots of different types of customers using the product. And, and, and so what we've gotten very focused on is actually how do we really help, you know, very specific types of users and workflows. And so, uh, PMs and product folks are actually one of the key ones, right? And so really going in and focusing on the B2B side of the business and doing a really great job of, like, the product development workflows that folks use on Bolt, like, that's been, been a key focal point for us. Um, so I think, yeah, like, our, our B2B revenue year over year is, like, 10x'd. Um, and so it's just, that's, you know, the, the fastest growing segment of revenue for the company. That's, that's where we've been focusing

  13. 10:34 – 10:54

    Pioneering the category, and the "vibe coding" wave

    1. ES

      all of our time.

    2. CV

      So as I think about the category that you, in a way, pioneered, these cloud IDEs then turned into some sort of A pro- pro- AI prototyping tools, then evolved into full-on vibe coding end-to-end tools, what, what is the current state, like, and what do you think this is, this is going?

    3. ES

      Yeah. Yeah, for, for vibe coding in general, you mean?

  14. 10:54 – 13:03

    Why prototypes have to become production code

    1. CV

      Yeah, because I see that a lot of these companies, uh, like in your category, they're just allowing users to prototype something, right? That looked cute. It was more of a demo. But then you evolved into a real solution that allows deployment into production, so it's not a toy anymore. This actually works, right? So first of all, I want to learn more about what, what's going on with this and then, like, where do you think this is going next?

    2. ES

      Yeah, for sure. I, I think, I think that a lot-- To me, I think a lot of the core use cases that, that you're describing that we see today, I think those are gonna continue to happen here. I think there, there's gonna be a lot more depth to the workflows on some of these things. So, for example, the idea that you wanna go and, like, as a product, uh, like a PM or a designer, like, I wanna be able to go rapidly iterate on how the experience should feel. It's actually useful to do that in an environment that is detached from your production environment 'cause it's way faster to just, you know, iterate, get that feedback, et cetera. But the problem, though, is if you're gonna do it outside of, like, your production code base, how do you actually marry that back into the production code base in a way that your engineering team, right, signs off on? And so this is stuff that we've been working on on our side of creating a very seamless connection of, like, using the exact production components, having a very seamless handoff flow to developers where it automatically pulls in your changes and can code it, et cetera. So I think, I think we're gonna see, like, more depth of, of integrations and workflows that, like, co- Really pave out those sorts of experiences there. On the flip side too, of what you're saying, like, you know, we have a lot of entrepreneurs that use our product to, like, build their businesses, and so enabling, uh, people to, you know, for... We, we have lots of, of entrepreneurs that are, you know, uh, generating millions of dollars, et cetera, on our product. Um, but how do you, like, further enable people to, to do that? Like, what are the types of applications that they, that they are trying to make or maybe that they would want to make, and making our agent the best in the world at doing those things? Um, and then they're, those are kind of intertwined 'cause, you know, what is a prototype but, you know, a, a proposal to, to ship functionality into your real production code base, and, and perhaps that production code base is in Bolt or perhaps it's not. But yeah, tho- I, I think tho- that's kind of how we're seeing these things

  15. 13:03 – 13:30

    The "ChatGPT vs. Anthropic" market framing

    1. ES

      evolve.

    2. CV

      Yeah, and I, and I see the, the competitive landscape evolving that way. I, I, I think the analogy here is in LLMs. ChatGPT got the first mover advantage and kind of grew within consumer, while then Anthropic grew within enterprise, and now of course they are converging. But you get, they kind of picked lanes at the beginning.

    3. ES

      Yep.

    4. CV

      In, in your world, the vibe coding world, and you see the other company like Replit, Lovable, and there's always new ones popping up like mushrooms, right? Like-

    5. ES

      Yeah, yeah

    6. CV

      ... so how do you position your product?

  16. 13:30 – 15:25

    Choosing a specific ICP instead of chasing everything

    1. ES

      It's a really good question, and, and this is, um, this has been, you know, over the past, like, over the past year, I think it was a year ago when we, when we really made the decision to really orient more towards people who were building products professionally at businesses, right? So versus there's a lot of people that are excited about AI. Uh, they... It's super cool. There's a lot of, like, consumers, general... I'll just call them, like, general consumers that are very excited about AI. Uh, maybe they've had an idea for a long time, et cetera, and like, and we have a lot of people that, that come to Bolt for that. But, um, you know, the, the way that I view it is, is, um, we want to build tools that help people that are, that are, uh, professional builders of products, that's PMs, designers, engineers, et cetera, to really raise the ceiling of, of what they can do and, and, and remove complexity of even if instead of having to use coding agents like in a CLI or whatever, um, how can we help, help you go from A to, A to Z as fast as possible? And, and so that's, I think that focus is actually, that's where, you know, when we, when we go into, you know, our, our conversations on the sales side, um, w- you know, obviously, you know, we, uh, all the other players, including the Frontier labs, you know, there's kind of bake-offs that happen. Um, but this is really where the importance, and just generally for startups, the importance of, of really going deep on, on a specific ICP and, and workflows is really matters, 'cause that's, that's where we win. [laughs] Because we, we just have a, a level of depth on, on certain workflows and capabilities, especially for product builders, that the other guys don't because they're, they're going very broad, right? So to your analogy-

    2. CV

      Right

    3. ES

      ... of like OpenAI, Anthropic, we're taking a very Anthropic type o- type approach to this where we're really trying to make certain types of users and, and workflows within B2B type usage insanely great. Uh, and, and we're, we're not, we're not just spraying, uh, hundreds of millions of dollars at, at everything, um, you know, uh, on the consumer side or whatever

  17. 15:25 – 16:01

    Growing the market vs. going deep

    1. ES

      have you.

    2. CV

      Yeah. I think that's an interesting dichotomy. On the one hand, you want to grow the, the market, the addressable market or the category by, by expanding, but at the same time you want to be remembered by something very specific that you do very, very well. Um, and I, I noticed that you recently, uh, launched Slides capability. Right now you can also create-

    3. ES

      Yep

    4. CV

      ... interactive slides. That's awesome. Uh, you're also launching templates to make it easier for people to build apps. But that also creates new pockets of competition, right? Like the Canvas of the world, the Gammas of the world. So now suddenly become, uh, I don't know, frenemies in a way because you probably-

    5. ES

      Yeah, yeah

    6. CV

      ... interact with them, right? So how are you thinking about that?

    7. ES

      Yeah, it was funny. I was

  18. 16:01 – 18:31

    Launching slides, and positioning each product

    1. ES

      on a podcast with, uh, one of the, the, the Gamma founders a week or two... It was the week we launched Slides, and, and he, he, he was like, "Yeah, yeah, I saw you guys launched a Slides thing," like live on the podcast. And, and I kind of forgot that we had launched, because for us it was, it was just, it was, it's like a fun project we made where we made it for ourselves, and it's not like we're not trying to get into the business of Slides, you know, like a, like where that's like the big thing we do. Um, but we, we ended up talking on the podcast, me and him, about this, where, uh, it kind of everyone's competing with everyone. What I said to him on, on, on that podcast, which I'll repeat here, is I don't think it's bad. Like for us, like we, we, we, I, I think we, we added some unique, uh, ideas into the mix as far as Slides go. Um, and, and I think that, and the reason we, we even launched it is it's just as a, as a, as a, a, a product. If you're, if you, if you're an innovative company, you're gonna have ideas and you want to put them out there, and if you've got the cycles to do it, you know, it's, it's fun. And it's, we open sourced it too. It's not like it's this closed source thing. But I, I look back at the dot com era. There was a lot of this sort of thinking going on where everyone was competing with everyone on everything and everyone was trying to do everything. Where like back then it was like portals, like Yahoo and AOL, like your portal, like, oh, we, it's like media is coming to the internet. We have to have media, like people, like the homepage of the internet, we, needs to be Yahoo or AOL, and like we're gonna do all these things. Oh yeah, search is just this tiny box up there, but it's the portal, right? Except that no, s- that tiny box i- is, it was the juggernaut. Like that was, that, that focal point, like Google was like, "All this other stuff, forget it. Just the input box. How do we make that thing insanely good and the best in the world?" And that, and that was, that was actually... It, it was underneath the other guys' noses, right? But they were so unfocused that, that they, they completely missed it. And, and that's, that to me is I think the era that we're in, I, I just, I look back historically and, and, and, and, and maybe this time is different, but, but to me, the comps, the comps don't track, right? The, the comps to me go, we're in the middle of this, and so it may seem like this time's different until, until it's clear that this time isn't different and certain companies break away because they actually focused, right?

    2. CV

      Yep.

    3. ES

      Um, and so that to me is kind of, and that's how we view our product strategy. And it's not to say that we won't do things that, uh, we won't release things that would be competitive with Gamma or whatever, but it's like, even on this podcast, it's like- I think Bolt sides is cool. It's open source. You can use it with any agent. On the flip side, I'm, I, I, you know, I don't know Agama's feature suite. I mean, they've, they've certainly got a ton of stuff that, like, [chuckles] we don't-

    4. CV

      Right

    5. ES

      ... that we probably never will, right? And that's cool. Like, that's why you would wanna use those guys or whatever. I don't know what those things are, but like, I'm, there's no... That's their entire product. Sur-

    6. CV

      Yeah

    7. ES

      ... surely there's, there's depth,

  19. 18:31 – 19:22

    Clear positioning across the product line

    1. ES

      right? So, um, yeah.

    2. CV

      And the positioning I think is also very clear in each of the products, right? Like, you're starting from a vibe coding solution for PMs and engineers, where the other one started more as a solution for marketers. Eventually, there's some points of conversion, but, like, that doesn't mean that you don't have your own identity. I've seen this movie play out even in the product tech category. It didn't even exist. Like, there were, I don't know, Optimize.me was doing A/B testing for marketers, and then you had, like, I don't know, designers using Photoshop, and suddenly you start seeing tools that are being created specifically for PMs. And back in the day, the whole winning strategy was you have to be hyper, hyper specific, like analytics, prototyping, uh, roadmapping, and now it's about a platform instead of portal.

    3. ES

      Yeah.

    4. CV

      Now the, the game is the all-in-one platform, and everybody's overlapping. At the same time, they're also integrating because they're assuming that a lot of these enterprises are probably using different point solutions, and the winner, in a way, needs to allow the data to flow.

  20. 19:22 – 20:23

    The Netscape "two ways to make money" quote

    1. ES

      Yeah. It, it reminds me this, the, the quote from the old Netscape CEO. Uh, I got his name, his name slips my mind, but, you know, he said that, you know, there's, there's only two ways to make money in the software business, uh, bundling and unbundling, right? And so it's... And, and, and these things go in phases, right, where, you know, they, they will, bundling will happen and then, uh, then, you know, unbundling. Streaming services are an ex-excellent example of this, right, where, um, Netflix, uh, you know, really rose to, to prominence, um, by, by bundling, right? It was like, okay, I don't have to buy all these subscriptions, and now, [chuckles] now, now there's a whole bunch of streaming services, and so everything's getting unbundled. And, you know, so it's... I think these things kind of go in phases, and, uh, typically the unbundling become-- it becomes clear that the unbundling is, is where the, where the value's gonna be once, once there is, uh, there's too much bundling going on, [chuckles] right? Like, that's, that's kind of the issue. Um, and, and that's, that's kind of... I, I think we're, I think we're approaching that in the market, um, if not already. I, my, my, yeah. I, I, I think in some senses we're already there on certain, uh, uh, parts of these things, but,

  21. 20:23 – 23:49

    The open-source bet and signing the open-weights letter

    1. ES

      but yeah.

    2. CV

      Another thing you mentioned briefly is, uh, open source, and I think that is a big deal here. Uh, you recently signed the open-weights letter created by Microsoft, and I know you connect with so many different models. I want to hear more about your take.

    3. ES

      Yeah. I think, so StackBlitz as a company, we've been, like, uh, deeply, deeply invested in open-source software, um, you know, for, uh, since we were founded. Um, and that's not, not even just lip service. I mean, we've put, um, substantial amounts of, of money, um, even when we were not profitable, uh, behind open source and, um, backed. Like, you know, there's a, a project called Vite, which is pretty much how the, everyone does web development these days. I mean, we were the earliest backer of that. We hired people onto our team to just work on that thing full-time in open source, so like, it, it, there, w- this is, um, you know... There's, there's there there as far as, uh, you know, our roots to this stuff. When you look at, like, the open-weights models, it's key that these things remain available and usable, um, to, to all companies and all people around the world, and, and, and there's kind of a number of reasons for this, right? But again, I look, I looked at past parallels maybe similarly match the situation, I think back to the '90s with, um, Windows versus Linux, and if the amount of innovation that has come... By the way, Windows is fine. I mean, Windows is used everywhere, right? I mean, it's, uh, the Windows is, is doing great, right? And, and, and they've added a lot of value to this world. Boy, Linux is, has been a, a workhorse, right, for, for how, how the, like, the hyperscalers that we have, like our ability to spin up micro VMs, blah, blah, blah, everything, right? And, and back then, these same things were being said where, "Hey, hey, Linux maybe should be outlawed," like crazy stuff. In retrospect, crazy stuff, but... And, and it reminds me a lot of now. And I, and I, so I think that if you, if you wanna fast-forward 10, 20 years and kind of look back, like where's the bulk of the world's innovation, what's it being powered by, I w- I would be surprised if, if open-weight models were not a, a key pillar of that story. And I think just geopolitically, I think it's a, a very bad idea for us to, you know, if we were to try and bi- which it does not appear we are doing this at this point, but if West- Western countries were to try and bifurcate the, the open-source models coming from non-Western states, right, the availability then within Western states and, and forcing use of frontier labs, um, that, that's not, that tends to not be a good recipe for how, at a national level, how countries get to actually play to win, right? Um, and you see a lot of great companies in the States that are, that are doing really incredible things with these open models, and that's... It's exciting to me because there's some companies I've, I've talked to recently that are doing really great research on how to, like, you know, how to optimize open-source models to run at a fraction of the cost. You know, like 90%. Like, you imagine running Kimi K3 at one-tenth the current price, right? And it's, it's possible to do. It's a possible problem. If, if not 90%, 80%, like, you know, off of, of what the list price is. But that's because they have access to the weights. [chuckles] You know? Like, if you don't have access to open model weights that, that people can use, no innovation happens, and that's, you know... If, if those guys crack that nut, I mean, that's gonna be transformative, right, for the level of ubiquitous access to these models and-

    4. CV

      Yep

    5. ES

      ... and, and the, and the degree to which people can affordably use these things, right? So anyway, so I look at it, to me it's, it, it, it's unfathomable that we would, that we would cut off that sort of innovation, right? But that's what would happen if, if open-weight models were, were not permitted to be, to be

  22. 23:49 – 24:10

    Why not participating in open source is the real risk

    1. ES

      used.

    2. CV

      The way I think about this is, uh, it's much riskier to not participate in the open source game than participating in it. And I think about your product as an application layer on top of a ton of models, right? And I kind of trust you to, in some cases, make the right decision to pick the models that are optimizing for my use case as well as for my price sensibility.

  23. 24:10 – 24:50

    Open-weights models, price, and getting aggressive

    1. ES

      Yep, exactly. Exactly. Yeah, and, and, and, and I, I think pri-price is a really important thing where, you know, the open-weights models are allowing us to give more value to our customers at a lower price. Right? And, uh, so, so it's just you're hurting, you're hurting customers. Um, you're hurting the people that are trying to use AI if these sorts of models can't be used, right? And of course, the ability for us to actually, you know, use our own data to better tune the, the open-weights models out there so that it better fits the workloads that our customers are doing. Obviously, Cursor has been the phenomenal example of this with their Composer models, but that's, that moves the needle. That, like, moves the needle. It's a g- a way better experience at a, at a fraction of the price.

  24. 24:50 – 25:53

    How Eric thinks about pricing his own product

    1. CV

      So how are you thinking about pricing for your own product?

    2. ES

      Good question. I mean, you know, I, I think for us, we wanna get really aggressive on the pricing. A- again, I think that the, the big sea change happening here is the open source models are really catching up with the labs, and there's, there's a very tiny, I wouldn't even call it last mile at this point, but it's like there's some, there's a, there's a last bit of distance to really kind of marry these things where we can provide the same level of experience, right? But, you know, this, we're, in, in, in a couple, some number of weeks from now, we're gonna be rolling out some stuff that will, could be our first foray into this, where, you know, we, we can, we can go and offer stuff at price points that just, it would not have been possible before while providing a great experience. But that, that to us is we're, we're all in, giving our users the, the most bang for buck. I mean, that's kind of from the get-go for us. That's always how we've approached this is like when we figure out ways where we can reduce costs, right, which is, you know, always a top concern for us. Um, uh, not just for us as a business, but it's actually really more like, hey, how can we actually make this a better economical, uh, decision for the folks using the product? Like, we, we are always trying to figure out how we can make this, um, better and more

  25. 25:53 – 26:14

    The billboard, the "SaaS apocalypse," and old pricing models

    1. ES

      affordable.

    2. CV

      Speaking of pricing and being aggressive, the other day I was driving around and I saw a billboard, uh-

    3. ES

      [laughs]

    4. CV

      ... built by your company that said, "Buy SaaS," or let me, let me find it.

    5. ES

      [laughs]

    6. CV

      "Sorry, SaaS." [laughs]

    7. ES

      Yeah, yeah.

    8. CV

      So I want to hear your take on the SaaSpocalypse, right, and how these, uh, previous pricing models are probably not going to cut it and what you are thinking about it to make sure that you end up

  26. 26:14 – 29:52

    The campaign, and getting the industry-defining calls right

    1. CV

      on the winning side.

    2. ES

      Yeah, yeah, for sure. Yeah, so we're doing this whole campaign on the, yeah, so we have all these billboards in like New York and, and San Francisco that say, "Sorry, SaaS." Um, we're like, we're apologizing for the SaaS apocalypse. But yeah, so I, I think, um, you know, I think the per-seat model thing, I think that's, that's the biggest risk point, I would say, for SaaS businesses, right, is with agents, as agents become more capable, more powerful, you may just, you may not need as many seats for certain types of products, not because, like, they're, those people won't be at the company, but where it's you're gonna have agents that are going and doing things, and it makes a lot more sense to charge based on usage and access and, and that sort of thing. So that, and that just across the industry. It was, when we launched Bolt, actually, it was at, it, uh, [laughs] I'm very proud of our team. We got a lot of things right that ended up being how the industry now works, where when we launched Bolt, at that time, ev- all these AI tools, they all had been doing the, they, they, for, you know, for whatever reason, people were charging like Netflix does, where it's like, hey, one price, all you can eat, except in the case of these AI things, if you eat too much, you, we're gonna ch- you know, kind of throttle you and make you come back later. But it's like 20 bucks a month. That was kind of like Copilot was doing that. I think Cursor was doing that. Like, all the, everyone was doing this, right? And when we launched Bolt, the demand was so nuts in the first 24 or 48 hours. People ripped through, w- our subscription was like at nine bucks at that time. People ripped through that in not even a day. [laughs] And they're like, "I wanna give you more money." And so what we ended up doing was within 72 hours of, of that, we shipped usage-based pricing, where you could actually choose, you know, you could like upgrade to the amount of usage that, that you wanted. And, um, it was, I think it was within a few weeks of that, uh, Barclays, um, you know, they're, they're like a investor banking sort of, you know, outfit or whatever, uh, mega. They, they wrote a report that they sent to, you know, their, their whole client base or whatever that said, "This, we think that this model, that this, this random company, StackBlitz Bolt," [laughs] like at the time, "We think this model is actually gonna end up being how, how this stuff gets priced." And that is exactly what has happened, right? Where all these AI products, everyone's charging based on usage. Um, you can choose how much, you know, for like in the case of like Claude, they have like the $20 plan, the $100 plan, the $200 plan. I mean, that, that's, that is what we pioneered, right? And so, so I look at things like that. I think that's just gonna continue, right? Where it's, okay, how do you actually, you know, how do you charge based on the value being provided versus just, oh, hey, you have 10,000 people at your company, and if you want them to be able to log in and use this, they're gonna have to pay. It's like, how much value are all those people getting out of this, right? Like, is it, is the value actually there? 'Cause if it's not, then what's probably gonna happen is people are gonna be using their agents. They're gonna go and, you know, hit your APIs, blah, blah, blah, and then, uh, there's gonna be three people at your company that actually need to use your product and 2,000 that are getting this, you know, through an agent via hitting the API. You know what I mean? So it's like that's, you, you have to kind of go, "Okay, how do we actually, uh, price to value?" So to me, it's actually kind of a, a healthy, how do I put this? Um, it's very, it's, uh, you know, to be in the, in the incumbent position, of course, is uncomfortable, but like such is the nature of free markets. The, the reason that those margins are under pressure is, is it's not because this stuff is bad, blah, blah, blah. Like, and what we're seeing is people are not losing their jobs because AI, and, and, and to at least certainly to the degree that some AI leaders have said it was gonna happen. What's actually going on though is it's like, it's bringing into question, like what is the value of that, of all these seats that you're forcing people to buy, right? And in a lot of cases, it was really more of, well, we have to do it.

    3. CV

      Yeah.

    4. ES

      Not because we're getting this much value. So that, that just kinda comes back to, okay, well then you need to go innovate, right? So that's, that's kind of, that's, that's my view on, on the

  27. 29:52 – 30:26

    The end of captive-audience B2B SaaS

    1. ES

      SaaS side of things.

    2. CV

      And we are seeing this across the board, right? Back in the day, I think, uh, B2B SaaS was mo- mostly a captive audience 'cause first of all, the person who's buying the thing is not using the thing. That is a, already a, a, a yellow flag, but also like usage was only being measured at the maybe the week before renewals [laughs] where like all these-

    3. ES

      [laughs]

    4. CV

      ... customer success managers will have to prove, "Hey, someone in the company did something very good, hence we deserve thir- 12 more months of grace," right?

    5. ES

      Yep.

    6. CV

      And now you're kind of on the hook, and you are sharing risk and reward, but ultimately, uh, the, the access to value and the access to usage is right there

  28. 30:26 – 31:32

    Fattening margins where value is delivered

    1. CV

      for everybody.

    2. ES

      Yep. Exactly. Yep. And so I think to me it's again, it's, I think it's just a, it's a good unfattening of margins where value was not tremendously present. 'Cause then you see other, th- there are SaaS companies that are crushing it, right? Like Shopify's crushing it, you know? The, and, and so I think earlier in the year there, there, you know, the, the, there was an over-rotation on, "Oh, SaaS is dead," and blah, blah, blah. And, and what's, what's, the, what's, what's really going on is, I mean, I think that there was pending question of, okay, well what of these things, are any of them, do they, the fitness function of them, are, are any of them gonna, is it clear any of them are gonna be able to thread the needle here? And, and, and now we're seeing the first crops of these where it's like, yeah, these guys, these guys are doing great. And, and so I think those are the gold standards of, of what the other SaaS companies have to look up to and go, "Okay, this is, this is how we have to approach the problem." And similar sea change that happened during the dot com era, right, where, you know, the traditional media, print companies, et cetera, like they, these sorts of questions, right? Some, some thread the needle really well, some did not, but it's again, it, uh, such is life in business. Um, how good are you at embracing pain, uh, quickly? [laughs] And, and, and, and, and embracing it and then pushing through it and figuring out what, what needs to

  29. 31:32 – 31:46

    Wrap-up

    1. ES

      be done, right?

    2. CV

      Such a beautiful way to wrap up this, this conversation. Thank you so much for your time, Eric. It's been a pleasure to learn from you.

    3. ES

      Yeah. Thank you for having me. This has been a blast. [outro music]

Episode duration: 31:46

In the uListen app, the transcript lights up word by word as you listen. Get uListen

Transcript of episode ZELi-yvOF_w

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.

iOSAndroidClaudeChatGPTChrome