深入 Bolt:从濒临死亡到历史上增长最快的产品之一 | Eric Simons
Inside Bolt: From near-death to one of the fastest-growing products in history | Eric Simons
Introducing the Guest
Lenny Rachitsky: The rate you’re growing is absurd. You’re in this cohort of companies that are just growing at rates that we’ve never seen in the history of startups.
Eric Simons: The company was on the verge of going under when we launched Bolt, and what ended up happening is, in the first two months it went from zero to 20 million of ARR. And we’ve already crossed 30 million of ARR, with the current rate we’re on, our forecast for the year is we want to get to 100 million of ARR.
What Exactly is Bolt
Lenny Rachitsky: This is just non-stop wild shit. How is this possible? What has allowed you to grow this much, this fast, with such a small team?
Eric Simons: Most importantly, it’s been the people. It’s rare to find startups where you have the core group of five, six, seven people that have been there for five years plus.
The Growth Metrics
Lenny Rachitsky: You basically were building a tech first, and then looking for a problem to solve later, which is often what people tell you not to do.
User Growth vs Team Size
Eric Simons: I think that’s the hard thing about being an entrepreneur. There are periods of time where you have to make judgment calls that are not going to be the consensus view. You got to have confidence in your convictions on how to best play the hand.
Lenny Rachitsky: A lot of people see these stats, and they sometimes don’t see that there was also years and years of work before that.
The Sustainability of Growth
Eric Simons: It was kind of like, Bolt’s this overnight success, seven years in the making.
Bolt Product Demo
Lenny Rachitsky: Today my guest is Eric Simons. Eric is co-founder and CEO of StackBlitz, which makes a product called Bolt, which is currently neck and neck with Cursor for being the fastest growing product in history. They’re currently the number one most popular web AI code app with over three million registered users. Two months after launching last October, they hit 20 million ARR. At the time of this recording, they’re approaching 40 million ARR. The story of Bolt is wild. They actually started the company seven years ago, and were about to run out of money and shut down. But they realized the tech that they’d been building for the past seven years, called WebContainer, was perfectly suited for building AI products in a browser. So they launched the product with a tweet, and as Eric describes it, it was an overnight success seven years in the making. If you’d like to better understand the cutting edge of AI coding apps, and where things are going with AI and product building, this episode is a must listen.
If you enjoy this podcast, don’t forget to subscribe and follow it in your favorite podcasting app or YouTube. Also, if you become a yearly subscriber of my newsletter, you now get a year free of Perplexity Pro, Notion, Linear Granola, and Superhuman. Check it out at lennysnewsletter.com.
With that, I bring you Eric Simons.
When I was at Airbnb, one of the things that I loved most was our experimentation platform where I could set up experiments, easily troubleshoot issues, and analyze performance all on my own. Eppo does all that and more, with advanced statistical methods that can help you shave weeks off experiment time, and accessible UI for diving deeper into performance, and out-of-the-box reporting that helps you avoid annoying, prolonged analytics cycles. Eppo also makes it easy for you to share experiment insights with your team, sparking new ideas for the A-B testing flywheel.
That’s the point. You can start investing in minutes, and with as little as $10 by visiting fundrise.com/lenny. Carefully consider the investment objectives, risks, charges, and expenses of the Fundrise Flagship Fund before investing. Find this information and more in the fund’s prospectus at fundrise.com/flagship. This is a paid ad.
Eric, thank you so much for being here and welcome to the podcast.
One-Click Production Deployment
Eric Simons: Thank you for having me. Yeah, I’m stoked to be here.
Lenny Rachitsky: For folks that are not super familiar with Bolt, what is Bolt?
Native Mobile App Development
Eric Simons: It’s really simple. You go there, there’s a text box, and you tell it what you want to build. Whether it’s a web or a mobile app, and so, it’s kind of one of these text to app building tools that have become pretty popular over the past few months here. And it’s not just building a static site, or something like that, but you can actually build full stack, real software with databases, and hosting and et cetera, just from prompting. And in a ridiculously short period of time, it’s not like you’re spending hours and hours or days, putting this together. You can get results in like, a minute.
Seven Years to $40M ARR
Lenny Rachitsky: Let’s just share some numbers about the scale of what you’re building. The rate you’re growing is absurd. You’re kind in this cohort of companies that are just growing at rates that we’ve never seen in the history of startups, and you guys are at the edge of that. Share some numbers about how things went when you launched, and where they’re at today.
Eric Simons: Yeah, when we launched, the company was on the verge of going under when we launched Bolt, our company StackBlitz. We’d been around for seven years building web-based development, environment stuff. And so when we launched this we were like, “This would be amazing if this added a 100K of ARR over the next couple of months.” And what ended up happening is, in the first two months, we went from zero to 20 million of ARR. And I think we’re on month four, or four and a half or something like that at this point, and we’ve already crossed 30 million of ARR, and we’re on the verge of crossing 40. By the time this comes out, it appears that we’re going to be at 40 million ARR. So it’s just, the scale of the growth of the revenue has been nuts.
And of course, that correlates with insane user growth, as well. We’ve added three million registered users just in the past few months here, and monthly active users is around a million, I think at this point, per month. So it’s just been… I’ve never seen anything, I’ve been doing startups for 15 years, I’ve never seen anything like this. Everyone I’ve talked to, our investors, or et cetera, there’s not a lot of corollaries to what’s going on here. And it’s kind of extraordinary, because our company wasn’t doing AI stuff six months ago. We had no AI products, and just out of nowhere we, from almost death of the company to being the number one buy traffic revenue, et cetera, like AI Cogen app, that’s totally web-based, in the world. I think the only other startup ahead of us is for Cogen, just in general, would be Cursor, and the option revenue at this point.
And so anyways, it’s been a heck of a ride. And our team’s like 15, 20 people, so it’s just dealing with, we’re going to be closing on 100,000 customers, and our support team’s like three people. So we’re trying to scale as fast as we can. So it’s just kind of mind-boggling, just the scale of the demand, and how we’ve had to turn things around to match the demand as best as we can.
Lessons from Deep Tech Bets
Lenny Rachitsky: Mind-boggling is an excellent way to describe what you just shared. A million monthly active users, you’re at 40 million annual recurring revenue, five months into the business. Is that right?
Startup Survival: Controlling Burn Rate
Eric Simons: Yeah, single digit. Yeah, single digit weeks. That’s the current track rate that we’re seeing for the thing. Yeah.
Lenny Rachitsky: I think, are you guys are the fastest growing startup in history?
The Founder’s Non-Consensus Bets
Eric Simons: I mean, I think it depends on probably where you peg the number. Because yeah, we’re here to just build great products, and just push the limits of what’s possible with the technology. And I think that we do our jobs well, kind of crazy things can happen, but I mean, the current track rate we have, we’re going to be exceeding the forecast for Q1 with the current rate we’re on, and our forecast for the year is we want to get to 100 million of ARR. And now, I think there’s been a couple, that would either be on par with Cursor, or ahead of them, or something like that.
And I think there’s going to be more things like this, too. I don’t think that… It just, there’s something really, I think a lot of people are in disbelief about it too, where they’re like, “This is, okay.” And this is from when we were at, got to four million ARR, five million ARR in the first month. I would talk with people and they’re like, “Okay, yeah, but that could go to zero.” And then it went to 20 the next, “Ah, it could go to zero,” but now we’re closing on 40.
So from my view, I was also very skeptical, as this. I’ve never seen anything grow like this, right? And so part of me, for like a month I was kind of waking up waiting for the day where it just was like, “Okay, it’s over.” You know what I mean? This crazy thing happened, and now it’s not. But that data just hasn’t come. And you see this happening with Cursor, you see this with a lot of these other AI startups. And the value proposition is real. The free market is filled with rational actors. People are coming to these tools because it is solving problems, they’re able to do way more for way less cost, than it would otherwise. And that’s why I said, I think we’re going to see more of this, whether it’s in coding or other verticals, or whatever. In a sense, it’s almost like maybe the new normal, as AI just continues to get better. But, anyways.
Lenny Rachitsky: Let’s get to a demo of Bolt, so people can actually see what this looks like in action. And as you go through it, if you can even point out stuff that is different from other products in the space, say Lovable, VZERO, Replit, that other folks have heard about, that’d be useful.
The Post-Launch Explosion
Eric Simons: Awesome. Cool. Yeah, so this is Bolt, you just go to bolt.new. Things that I think are really interesting about Bolt. One is, it’s just dead simple. Whether you’re logged in or logged out, it’s the same UI, it’s extremely simple, it’s just a text box. And I think that the biggest difference between Bolt and the other stuff out there, it’s actually subtle. It’s not like something you’d necessarily see in the UI, but it’s how fast it is, and how reliable it is.
And this is because of how we are actually doing the compute, because what’s going on here is when you type into, whether it’s Bolt or another product, it has to spin up a dev environment to actually make that application. So there needs to be some operating system somewhere that’s running it. Everyone else runs those things on cloud servers, which those can take minutes to boot up, and they often will run into issues, and then you can end up literally stuck and have to contact support to get it done, and get it unstuck.
With Bolt, and for the past seven years, what our company’s been doing has been building an operating system that runs inside of your browser locally, using your CPU. So we have a very permissive free tier, and it’s insanely fast, and it’s insanely reliable.
So if I want to, just as a quick example of this say, “Make a clone of Spotify,” and just hit enter. This thing’s already getting to work, and already, on the right here, this is a full dev environment. This is an actual operating system, running inside of my browser. And I can run commands on it, et cetera. And really, what you’re seeing down here, this terminal and kind of what’s backing it, this is what took us really five, six, seven years to build, and make so reliable. There would not be a Bolt without this technology called WebContainer, that allows us to run an operating system in the browser.
Because what’s going on here is, our AI agent for Bolt has bidirectional communication with this operating system. It’s writing code, it’s running the dev server for this thing, it’s going to go ahead and spin this up. You can see how fast this is, in a matter of 60 seconds I said, “Make me a Spotify clone,” and now we have one. And it looks pretty darn good.
20 People Handling Explosive Growth
Lenny Rachitsky: That looks really good.
Team Trust and Cohesion
Eric Simons: And that’s one of the other aspects around Bolt is, this technology we made for the operating system side, the guys that have been working with us for the past five-plus years on it, before this they were actually doing machine learning AI stuff. And so when it came time to write the agent for Bolt, we had just an incredible amount of in-house expertise on how to actually merge these two different technology sets, to have this really reliable experience that produces really beautiful, really functional stuff. So that’s based on what’s really cool about the Bolt experience.
The other thing is, a lot of these products it’s like, you can make something, but often you want to actually have a URL where you can share this. Having, maybe even attach a domain to it, or whatever have you. So with Bolt, we actually have built-in integrations with production grade hosting providers like Netlify, and for databases with Supabase.
So if I go and just click the deploy button here, this is actually going to run a production build of this project we made here. And again, this is doing this entirely inside of my browser, so it doesn’t cost us anything to do this. So again, you can do this for free, and it has gone ahead and deployed this on a real URL, on Netlify. This is live, I can share this with anyone, and if I want to buy the domain spotifyclone.com, and point it at this, I can click this link here. That will kick me into Netlify, I can attach this to my account, buy a domain, point at that thing. And then from there on out, whenever I’m prompting Bolt to make changes to this application and hit deploy, that goes live on my public website.
So this is the simplest way to build a web app that’s ever existed. That was one of the key realizations I had, a couple of weeks into the thing, I was seeing people use this for personal use cases. Like medical donation sites, or weddings, or whatever. And I was like, “Don’t these people know that Wix or Squarespace exists? Should I tell them?” And then it hit me. Those things are so complicated to use. I don’t know if you’ve ever seen just the UI of these things, but they’re crazy complicated, and that’s just for building a static website. There’s no way you could actually build a functional app. And that’s like, with Bolt, if we were to sit here for another 30 minutes, we would have streaming. You’d be able to make playlists of different MP3 files, or whatever. You can just keep prompting this thing to keep adding functionality.
So that’s, I think, some of the cool core experience of both here. I can show you something cool that we just launched, if that would be of interest.
Lenny Rachitsky: Let’s do it.
The Hiring Philosophy
Eric Simons: So this is like web apps, right? Web apps are amazing, but often you want to have a native app. And it’s hard to build web apps, it’s even harder to build native apps, that can actually, that you can then go put in the app store. And so we partnered up with a company called Expo, and their entire business is making, basically, React Native tooling and this ecosystem that makes it super easy to build beautiful apps, and actually get them in the app store.
And so right here, I’ll zoom in a little bit, we have this little, “Build a mobile app with Expo.” So if you click that, we kind of instruct you on how to just prompt mobile apps into existence. So yeah, let’s make another Spotify clone that’s an actual native mobile app. Let’s say, “Make a Spotify clone,” go ahead and hit enter. And what this thing’s going to do is actually, again, spin up a operating system here, where it’s going to boot up the Expo tool chain and actually go and make a mobile app for us.
And what’s cool about this is, we could actually preview it just in the browser here, but once this thing’s done and it boots up, it’s going to show a QR code, we’re going to be able to scan it, and in real time actually basically have a test flight of this native application that we can try it on our phones, and as we keep prompting you’ll see it making changes and stuff. This is kind of the first time that, you don’t have to be technical to make production grade web, full-stack web and mobile apps. At this point I’ve done nothing that requires developer knowledge to do any of this stuff.
And I think that’s what a lot of people are really excited about with this, and you know, the majority of our audience are people that are not developers, that are using this. They’re PMs, they’re designers, they’re entrepreneurs. Because these are people that have always been great at building products, but previously, the only way that they could get their ideas into coded software was through a developer’s fingertips. And now, they can deal with their own, through prompting.
So you can see here, we’ve got this little QR code. I’m going to go ahead and scan the thing.
Bolt’s Priority Decision Making
Lenny Rachitsky: I’m going to do it, too.
Tech Stack and Product Workflow
Eric Simons: Cool.
Building Real Products with Bolt
Lenny Rachitsky: By the way, I love that you had just enough things to say until it finished. That was pro.
The Limits of Existing Codebases
Eric Simons: Just as I planned, you know? So on my screen it’s booting up, it’s bundling the JavaScript of this thing, and it’s beta. We just launched this last week, by the way. So if you can kind of see on my screen here, I actually have this Spotify looking app, right?
Reevaluating the Value of PMs
Lenny Rachitsky: Wow.
Eric Simons: That, you know.
Organizational Structure and Change
Lenny Rachitsky: That looks like, exactly like Spotify.
Eric Simons: It looks exactly like Spotify, right?
Advice for the Next Generation
Lenny Rachitsky: It’s good.
Eric Simons: Yeah.
Most Important Skill in AI
Lenny Rachitsky: We’re going to be sued right now, so let’s be… You’re doing too good a job with this. No, that’s amazing.
Eric Simons: Yeah. So it’s pretty cool, right? And so what’s cool is that, and as you keep prompting on your device, it’ll just keep reloading. Without you having to kill the app, you can actually see the functionality getting added. And so, in this use case that you and I have right now, it’s like if you and I were building an app together, we could be on other sides of the planet and you could actually be not just seeing a screenshot of the thing, but actually touching it and feeling it, and putting it through its paces.
And so a lot of product teams, I mean, this is just changing how people do product development. It’s faster to do this than design a whole bunch of Figma frames, necessarily. Right? So.
AI First, Humans Come Second
Lenny Rachitsky: We’re going to spend a lot of time on that. Okay, this is incredible, this whole episode so far is you just blowing my mind and I imagine listeners’ minds, just over and over and over. I don’t even know where to go with all this, sometimes.
You made a really important point, that you worked on this for seven years before you launched Bolt. A lot of people see these stats, zero to 40 million ARR in five-ish months, and they sometimes don’t see that there was also years and years of work before that. And the reason that you guys have been so successful is all the work you did that allowed, that built this WebContainer technology, it sounds like. Is there anything there that’s worth sharing, you think, of just that part of the journey? I know we’ll go through the origin that all, where Bolt came from, but I guess just that WebContainer component specifically. That feels like a huge deal.
Why Sonnet Changed the Game
Eric Simons: A hundred percent it is, yeah. And I would say this is, surprisingly to me, it’s still one of the contrarian viewpoints of our company. Because over the years it was like, when we first… And that, the WebContainer was the bet, that we made the company on. Just to be clear. StackBlitz was a browser-based, deep technology play on, “Can we make a web assembly based operating system that can boot in a browser, in like a hundred milliseconds, and run full on development tool chains?” That was really it.
And we’d gotten the idea for this, and the insight that this might be possible, because back when my co-founder and I came out to the Valley, he and I grew up down the street from each other in Chicago, we wrote code together at 13, and been building stuff ever since. And we came out to the Valley in 2012, and we just had the good fortune of bumping into Dylan Field and Evan Wallace when they were building Figma, in the early days. And that was, I don’t think a lot of people know that Figma was also a browser-based deep technology play. Their first pitch for Figma, they didn’t have a design tool. Their first pitch was this 3D ball dropping into water, inside of a browser town.
And the pitch basically was, “Browsers have this new capability called WebGL,” the predecessor to WebAssembly, “and with these things, for the first time, you could actually create a graphics rendering engine, that you could then build a design tool on top of. But you’re going to have to write that rendering engine from scratch, because nothing exists that can just compile into WebGL, or whatever. And if you want the performance you need, et cetera, it’s going to take us years to do, but if we do it, we think this will change everything for design.”
And obviously, we know how that story has panned out now. And back in 2017, 2016, 2017, Albert, my co-founder and I, saw the same sort of story begin to play out, but for web development and development environments. And specifically there was some stuff that landed in browsers like WebAssembly, shared memory, service workers, these different APIs. And we were like, “Oh, wow. It should be possible, theoretically, to write an operating system in WebAssembly that could run Node.js, and NPM and all the tool chains on top of it, that you need to do web development.”
And that would be huge, because setting up developer environments, it’s a pain for beginners. A lot of people churn out. The first thing you do when you learn how to code is not even learning how to code, it’s how to set up your computer to even start writing the code. If you go join Netflix, or any of these other fan companies, the first month or two is you being onboarded, to run that stuff on your computer and set up your environment. And we’re like, “If we could just have that be something, you click a link and it just boots in your browser, that’d be huge.”
It’s also, if you look at the other productivity apps that have really worked on the web, they’ve all had this compute model, right? Figma, when you open a Figma document, there’s not like some cloud VM that gets spun up for you to render the documents. You’re dragging things around. It’s using your CPU and your memory to do the work. Same thing with Google Docs. That’s the only model that’s ever scaled to a billion users. And so, when you look at Cloud IDEs, like Cloud 9 was the first one, back in 2009 or so. The way these have always worked is that your browser’s basically doing nothing, when you go to that. Every user that gets connected, there has to be a cloud VM that gets spun up for them, and then your browser’s just taking your keystrokes, sending it to the server, and then sending back the results of it. And that’s how all these other AI code, text to app sort of tools work. They’re all using cloud VMs.
And the problem is, on a small scale it can work, but as you scale it up, I mean there’s not even a 100 million VMs to rent, on the planet. But there are a billion devices that you can run this stuff on. Because that’s kind of what we’ve seen with Bolt where, if you want to build a product that’s going to be able to scale to that size, you have to look at all factors and go, “We have to build, make sure the technology provides the best experience, zero latency, transient cost.” There’s a permissive free tier, because the other problem with the server is, you end up, if you have a free tier, people are mining Bitcoin on it, they’re DDoSing people using your servers. So inevitably, you have to nerf these things and roll them back. But if it’s all done on the end device, it doesn’t matter.
So anyways, WebContainer was the key piece, and what we struggled with, it took us four or five years or something, to build WebContainer. What we struggled with for the years after that was just how to build a product around it, because developers loved it, but they weren’t using it in ways that they would pay money for. And as much as the nerd side of me wished that that would be enough, that it was like, building cool technology was enough. It’s like, “It’s not. We’re here to build a venture scale company.” And so that was kind of why we were high at the end of the journey, where it was like, we’re taking shots on goal. And at some point, this got a connected bat, right?
Model Improvements Are Just Starting
Lenny Rachitsky: There’s a lot of really interesting lessons from this journey, that I think are counterintuitive. One is, you basically were building a tech first, and then looking for a problem to solve later. Which is often what people tell you not to do. And it worked out, in this case.
The other interesting takeaway here is, it feels like it’s a similar moment to when AJAX came out and then everyone’s just like, “Wow, you can build new things here.” So it feels like there’s a lesson here of just, “If there’s a new technology that has enabled, something big that we think may, let’s just work there for a while, and see if something comes up.”
And then I think the other lesson here is just, as a founder, just survive as long as you can. Because you may find something that works.
Eric Simons: All great points, all great points. Because you’re dead right. And fortunately, my co-founder and I had, we had built a lot of unsuccessful stars before this. We spend most of the, or 20 times, churning through ideas on things. So when we had conviction, I was like, “This seems like a technology that will be important.” It seems like, the web is the most ubiquitous… The pitch or the theory in our head was like, “The web is the most ubiquitous platform in the world, but yet it has no, you can’t use the web to build the web.”
Every other platform, Mac has Xcode. Windows has Visual Studio. The web had nothing. And we were like, “At a minimum, Google should probably buy this thing from us. It seems like it should probably be part of Chrome,” at a minimum. And we thought, “Hey, this could be a huge enabler.” The vision of just making it as easy to build full stack applications as using Canva, it just seemed really compelling.
But when you do that sort of risky deep technology play, you need to… And we were very good about this, like the previous company Albert and I did, we bootstrapped it all the way through to acquisition, so we understood and we were living hand-to-mouth, to bootstrap that thing. So we understood out of it how to have a low burn rate, and take a lot of shots on goal, and make every dollar stretch beyond what anyone would think is reasonable or possible. And that’s how we played our hands with StackBlitz. We didn’t raise money for the first two or three years of the company’s life. We were bootstrapping it. When we did raise money, we barely spent it. Largely because it was like, “We need to just take a lot of smart bets, and it doesn’t make sense.”
And I would just say generally, until you see pull, just people pulling the product out of your hands, you don’t want to be spending money. You should be like, default, no. And when you go and buy software, you should be going, “We’re a tiny startup. Can you sell it for half?” Everything you buy, just keep the burn rate as low as possible, because you need as many shots on goal as you can possibly get. Because you have no idea. I think just generally, for startups, that’s the right way in my view, to approach it. Unless you’re seeing, again, immediate demand and pull, or whatever.
But yeah, I think that’d be, maybe the extra context I’d add on top is, I think that we ended up doing a good job of being extremely conservative. During a time in which, during 2020, through 2020 and 2021, which were times where exuberance and growing headcount was like, KPIs of companies. And were things that were being… With lot of emotional force of like, “Hey, you guys ought to be doing this.” And I’m glad that we didn’t heed the advice, because if we had tripled the company and kicked up the burn rate, there would be no Bolt. We would’ve gone out of business a lot of time ago.
So I think that’s the hard thing about being an entrepreneur, I think is you kind of have to… There are periods of time where you have to make judgment calls that are not going to be the consensus view. Maybe years later, it’ll become the consensus view, but you got to have confidence in your convictions on how to best play the hand.
The Next Steps for Bolt
Lenny Rachitsky: There’s so many great lessons here, I think just this idea of just staying alive. Dalton came on the podcast, he’s a partner at YC Ones, and he just had this phrase, “Just don’t die.” And that’s exactly what you guys did, seven years of just trying it until something worked, and I love that you actually were planning to shut down the company right before you launched Bolt. And I know you launched it with just a tweet, right? That was the launch moment?
Eric Simons: Yep. Yeah.
From Figma to Full-Stack Apps
Lenny Rachitsky: Maybe talk about that moment of just, after launch, signs that, “Okay, this is working. Something’s different.”
AI Developers Inside Slack
Eric Simons: Yeah, yeah. So day one it was like, there’s great reception to the tweet. We were like, “Wow, this is one of the biggest things, launch day reception we’ve ever seen.” And I think on the first day, I think we added 60K of ARR, or something. Which was like, I mean, crazy. Again, we were at 600, so we added 10% in a day. And I remember our dev ops engineer, he was the one who would flag me. He was like, “Guys, we got 60K today. This is crazy.” And I was like, “Yeah, yeah. But this is launch day.”
There’s the tech crunch, peak of initiation, in the classic startup-
Lenny Rachitsky: [Inaudible 00:30:17]-
The Engineering Team Tech Stack
Eric Simons: … star, yeah. I was like, “Listen, guys.” I’m trying to temper enthusiasm for the team. I’m like, “This is great. Got a lot of work to do.” And then the next day we added 80K, or whatever it was, and it just kind of kept going. And all the while, the product we put out, we built a thing in 90 days. We built Bolt in 90. So there’s a lot of things that were missing in the product. Like, basic stuff, basic stuff. And which, again, we cut the right corners on the thing to get it online, but we had this just growing influx of people using it, going, “How is there not a mobile responsive view? How are chat messages not,” we got to 20 million of ARR without a mobile responsive view, by the way. Just throwing that out there. It was like the iPhone not having copy and paste until iPhone 5, or whatever. That was that, this was that for us, it was like, no mobile. You looked at it on mobile, it was terrible.
But there was stuff like that, so we had to just… And then, we’re a small team and so, we were completely unprepared for just the growing traffic. And there was a whole bunch, I mean, the list of problems that were happening every single day was nuts. I mean, to start, we had never had a plan on stackblitz.com for more than $9.00. We had one price, nine bucks. And so when we launched Bolt we were like, “Again, we don’t think, hopefully people like this, but nine bucks doesn’t get you a lot of inference.” And so people burn through nine bucks in 48 hours. And they’re like, “I want to buy more. How do I buy more? Why won’t you take my money?”
So it was like, within the week we rolled out just completely new pricing plans, where you could upgrade, which ended up, has kind of now become the standard. All the other guys in the space have copied this. Where prior to Bolt going online, Copilot, all these previous AI things, everyone wanted this Netflix model where there’s one price, it’s like all you can eat, or whatever. And the problem is, if you do that, you want the inference cost to be kind of low, because you’re expecting people to use it a lot. And so you can’t do these agentic experience things, it would be too expensive.
And what we ended up stumbling into is that, “Okay, actually, people are willing to pay more. People want to pay for more inference, because we’ve crossed this threshold where you can get a very tangible ROI.” You know that this is providing a tremendous amount of value to you. So anyways, that was one thing, and the servers were just melting. Anthropic ran out of GPUs for us. Dario emailed me, he was like, “Listen, we don’t have anything more to give you.” At the times, where we’re like, “How do we deal with…” It was just bananas, for weeks. It felt like in 300, when they’re surrounded by 10,000 people, and our team is just doing everything. There’s 15, 20 people, just doing everything. My chief of staff and I were doing customer support 95% of the day. Anyways.
So yeah, it was a crazy wild time. I mean, it still is. We’ve had a little bit more time to grow into this. And usually, I mean as a company, to grow into even 20 million ARR, you get a year at least or something, to kind of staff up.
Important Advice for New Users
Lenny Rachitsky: Often, decades.
Direct Advice for Product Managers
Eric Simons: Yeah. So that was as hard, we’d go to people and kind of be like, “What do we do?” And the playbooks we get back are, take six months, or a year, or something. It’s like, “This isn’t going to work.” Which is funny, this is what it’s all about. I mean this is, at least for me, that level of intensity, it’s challenging. Fun challenges, you know?
The Famous AOL Office Story
Lenny Rachitsky: Wow, okay. This is just nonstop wild, wild shit. So you mentioned that your team was about 20 people through all of this, you guys are growing at this insane rate, 20 people. How was this possible? What has allowed you to grow this much, this fast, with such a small team? And this 300 visual is interesting, I imagine having these Spartans is a big part of it. Just what has allowed you to do this?
Eric Simons: Yeah, I think a lot of it again, I mean if you kind of look at where a lot of the other folks, like the Cogen types of app space, have really been struggling, a lot of it has been scaling their servers with stuff. And it was kind of like both this overnight success, seven years in the making. All of this stuff, there’s no way, if you rewind to year two, there’s no way we could have, we would not be at the growth on DAUs, and revenue, or whatever. There’s just no way. And so a lot of it is the technology we made, and most importantly, it’s been the people.
The people… It’s rare to find startups where you have the core group of five, six, seven people, that have been there for five years, plus. That’s a pretty rare thing to see in Silicon Valley. It is usually, folks are at a startup for a year or two, they kind of go to another one. You know what I mean? And the problem with the turnover like that is that you can’t take really long bets like the one we did. And so we’ve had, kind of from the get-go, again, this comes back from bootstrapping the previous company. Just having less people, and more context per head. That’s just been how we do it, and we feel very strongly about it.
And the reason for that is, one, that you can have high levels of trust with anyone you’re talking to, because you know that they have a lot of context. It’s not like this person’s completely in the dark, in some corner of the company that doesn’t… You know what I mean? The second thing, everyone has agency to actually get stuff done, front to back. And there’s no political community to get stuff approved by, there’s no… So when you look at what happened with Bolt, I mean, we had engineers that were, front to back, were on a call with someone running into an issue, going and fixing it. Cooking up the UI on the spot, and landing this thing. Without involving anyone else on the team.
So I think it was the culmination of just high trust, and people, we all just have enjoyed working together in the past. Maybe that’s why, that’s the only reason that anyone would ever stay at a company for that long, or whatever. And so those sorts of stressful situations, I think, are make or break. Those are make or break for any team. And so, I think that what’s happened is really, it is a direct reflection of the strength and the bonds of the people that are making this thing, and supporting the thing.
Final Thoughts and Outro
Lenny Rachitsky: Yeah, I think that’s such an important point, that you guys have been working together for many years. Most people won’t have that benefit. When you’re hiring people, when you hire this initial team, is there anything you look for that you think maybe people aren’t looking for enough? Anything you prioritize when you’re hiring new folks? Is it this idea that they can do a lot? They can do customer calls, they can do design, they can do engineering?
Eric Simons: Yeah, for us, and even if the folks were hiring us, hiring people that don’t care about the titles, and they don’t care about… It’s not like they’re… People, of course it’s people have a career trajectory, and that sort of thing, but they really are motivated by just working on cool things, and are chucking their ego at the door. And they’re there to collectively build something great, not just kind of follow, and be the brilliant jerk. Most of the people that we’ve hired have been in Europe. We’re a fully remote company. My co-founder and I are in the Bay Area. It’s funny, back in 2018, we rented an office and stuff, and we were commuting into it. Because we thought we’d hire people here, and like a year into it we were like, “What are we doing? You and I are coming to an office for 10 people, we’ve hired, the people working for us are in Europe, or across the US.” And we have one or two other people we’ve hired that are in the Bay Area at this point.
But yeah, I think we kind of look for folks that are intrinsically just trying to build great stuff, and are interested. And then the first people that we hired, the reason that we found them is that they were users of StackBlitz. A lot of people, the majority of people we’ve hired at the company have been people that actually came from our community, basically. So when we want to hire people, we put out a tweet and say, “Hey, we’re hiring an engineer,” and then we get DMs or whatever.
But yeah, those are the general kind of qualities we look for, though.
Lenny Rachitsky: I’m excited to chat with Christina Gilbert, the founder of OneSchema, one of our long-time podcast sponsors. Hi, Christina.
Christina Gilbert: Yes. Thank you for having me on, Lenny.
Lenny Rachitsky: What is the latest with OneSchema? I know you now work with some of my favorite companies like Ramp, Vanta, Scale and Watershed. I heard that you just launched a new product to help product teams import CSVs from especially tricky systems like ERPs?
Christina Gilbert: Yes, so we just launched OneSchema FileFeeds, which allows you to build an integration with any system in 15 minutes, as long as you can export a CSV to an SFTP folder. We see our customers all the time getting stuck with hacks, and workarounds, and the product teams that we work with don’t have to turn down prospects because their systems are too hard to integrate with. We allow our customers to offer thousands of integrations without involving their engineering team at all.
Lenny Rachitsky: I can tell you that if my team had to build integrations like this, how nice would it be to be able to take this off my roadmap, and instead use something like OneSchema. And not just to build it, but also to maintain it forever.
Christina Gilbert: Absolutely, Lenny. We’ve heard so many horror stories of multi-day outages, from even just a handful of bad records. We are laser focused on integration reliability to help teams end all of those distractions that come up with integrations. We have a built-in validation layer that stops any bad data from entering your system, and OneSchema will notify your team immediately of any data that looks incorrect.
Lenny Rachitsky: I know that importing incorrect data can cause all kinds of pain for your customers, and quickly lose their trust. Christina, thank you for joining us, and if you want to learn more head on over to oneschema.co, that’s oneschema.co.
I want to ask a couple more questions about Bolt, and then I want to zoom out, and talk about where things are heading in the future. Let’s talk about prioritization. I imagine you guys are just barraged with, as you described, after you launched, you’re just barraged with requests. Like you said, there’s a million monthly active users. I can’t even imagine the feature requests you guys are getting, plus all the stuff you know want to build. Just how do you go about deciding what to prioritize, and what to actually build?
Eric Simons: There’s a lot of things that you just don’t even know are possible to do, and so, people aren’t going to be necessarily explicitly asking for them. And so there’s been kind of a couple of these, where we use our gut instinct on like, “Hey, no one’s asking for this, in meaningful numbers at least. But we think this is going to be a big deal.”
Best example was last week with native mobile app support. That’s, by reception, the biggest thing we’ve ever launched. And it was something that, even internally at the company, some folks were like, “This, I don’t know. People are yelling about these other things.” And it is, it’s always this balance of how much are we just triaging various things, versus that new capabilities, but it was like, “This strikes me as an important one,” where we put some chips into the middle of the table on. And had it dead right. It’s just this mind-blowing experience, and now, there’s just thousands of mobile apps being created a day, that weren’t before. And how does that change things? I mean, now there’s small businesses that, they would’ve never made an iPhone app before. It made no sense. It’s super expensive. Now, that’s not the case.
So there’s kind of these things where it’s like, “Hey, we should go and take bets here.” But there’s kind of this, I think the best analogy would be like, it’s kind of like working at a restaurant, being like a chef. There’s some amount of, there’s feedback from the customers of, “This thing didn’t taste good.” And then there’s like, “Hey, we’ve been cooking something interesting, and this tastes… I don’t know. This, I think you’re going to like this. I think this is a killer dish.” And so you kind of have to balance those things.
And I think it’s actually, largely, a function of just years of experience doing it. I think if you kind of rewound 10 years ago, I would have had really no, I wouldn’t have had just the years of getting my butt kicked by the free market, to have cultivated a sense of this stuff. You kind of have to build your own gun and stick for it, I guess is the best way of putting it.
Lenny Rachitsky: To unpack this a little bit further, do you have a cadence you guys work through to decide what to build, and ship? Do you have a weekly meeting every week? Because I know the answer is probably, really, “It’s just chaos constantly, and fires we’re putting out constantly.” I know that’s a lot of it, but is there some kind of process that you guys have for deciding what to build, and how to share it, and just work with the team?
Eric Simons: We all meet every day. Pretty much the entire team gets on a call, and we just kind of front to back-
Lenny Rachitsky: You meet, like a Zoom?
Eric Simons: Yeah, every day at 8:00 AM Pacific, we’re on a Zoom for at least an hour-
Lenny Rachitsky: Every day? The whole company?
Eric Simons: Pretty much the entire company. Yeah.
Lenny Rachitsky: Wow. For an hour. Okay.
Eric Simons: Yeah. And we just go over everything and I think we’re going to probably start, as the team’s kind of growing, we’re going to start splintering off into different syncs, or whatever. But the thing about just having everyone in the same room every day is that, a lot of people will complain that it’s… On Twitter, you’ll see people say, “Oh, it’s the most expensive use of everyone’s time,” but it’s like, “Yep. But there’s 0% fidelity loss in that. Everything, every day, is being audited front to back, and being discussed front to back.”
So when you’re in these times of just extreme growth, you want as close to 0% loss, on communications. And so that’s how we’ve been doing it, especially since Bolt went online, and I think it was the week after Bolt went online we were like, “Every day, until we’re through this or whatever, we’re all getting on a phone call every day, and we’re front to back doing this.” And again, another reason why more context and less heads, every person at the company is aware of everything else going on at the company. So people can independently be making decisions that are generally, by default, more often correct than not.
Lenny Rachitsky: That is so interesting. I’ve never heard that before. Especially for company growing, that is like yours. That is super interesting, that that’s what you do.
Eric Simons: I don’t think we’re going to do that forever.
Lenny Rachitsky: Yeah, yeah. Of course.
Eric Simons: But, yeah.
Lenny Rachitsky: No, but I think that’s a really cool thing to note, that that works. And has worked for you. Where do you, so say you talk about stuff, then where do you put stuff? Where do you put your roadmap? Where do you plan? Just, what tools are kind of in the stack of the company’s toolset?
Eric Simons: Yeah, on the engineering side, we use Linear heavily. On kind of product roadmapping, we’re using Notion, and kind of making PRD type stuff in Notion. And we use Figma for design. No, actually, we use Bolt for a lot of design and prototyping at this point, as you can imagine. But yeah, I think that the tooling is nothing crazy. There’s nothing crazy sophisticated. I think we’ll be investing a lot more, and especially as you start splintering people out of being on the same call every day, so that’s where this stuff really starts to matter. Because you don’t have a time where you’re able to dynamically catch things that weren’t going to be brought up.
Lenny Rachitsky: I love that you guys use PRDs. I love that you even used that term. There’s a lot of talk of just like, “Oh, we got Bolt now we got all these tools, we don’t need PRDs. We’re just going to create a prototype immediately, and that’s it.” Talk about just why you still find that useful, and just what you put into your PRD, whatever that is for you.
Eric Simons: Unless there’s something that’s very sophisticated that we’re working on, we tend to keep them pretty light. I like to just have the minimal amount of context possible, that just ensures everyone’s on the same page and that the key outcomes for whatever feature that we’re working on, are going to be present when we get there. Because the things that, when these arguments get really beefy, you’re looking at it at, “God, there’s so much stuff to decipher.” The problem is, a lot of people are going to gloss over it when it gets kicked to development, or design, or whatever. It’s just going to start snowballing into a lot of stuff. It’s just better to keep it as simple as you possibly can. At least, that’s our approach to the thing. And often it’s some of these things are like, “Here’s a link to a Bolt.”
Lenny Rachitsky: “And here’s what it might look like.”
Eric Simons: Yeah. And not just look like, “Here’s kind of a working demo of what it will effectively feel like.” And then, be. Because that just, if a picture is worth a thousand words, a live actual demo is worth millions. You can feel it. It’s real. And that’s what we’re seeing, a lot of the businesses that are adopting Bolt now, that’s the use case that they’re using this for. Is high fidelity prototyping, because it’s now faster to make real prototypes using Bolt. Before, it was too expensive. The idea of, “And let’s prototype it, the engineers’ code a proto…” It’s like, that, it would take forever. It would be expensive. And now it’s faster to do this with Bolt, in code, and have a real working software product, than dragging around frames and Figma to actually make a static version of it.
Lenny Rachitsky: So let’s actually talk about that. Just how far have companies gotten with Bolt? Prototypes is where everyone’s kind of imagining these tools are at. I know that the goal isn’t just to make prototypes, it’s to build full scale. I imagine, long-term, Salesforce, Atlassian style companies, at scale. What are some examples of products people have built with Bolt that maybe would surprise people, in just how far they’ve gotten?
Eric Simons: Yeah, I mean, especially… When you’re starting greenfield stuff, you can use Bolt to build… you know. Like Salesforce, as an example. One of the first people that signed up for Bolt was this guy named Paul, and he’s an entrepreneur, and doesn’t know how to code. Built a CRM in three weeks, that has AI built into it and Stripe for billing, et cetera. He had gotten a quote from an agency for this, it was going to be 30 grand, and take six months. He had done in three weeks, and I think he spent 300 bucks on Bolt for the thing. So it’s like, this is… And he’s making money off of this. This is his start. Right?
Lenny Rachitsky: Okay, so he built this, and he’s selling it. People are paying to use it.
Eric Simons: Yeah. Yeah.
Lenny Rachitsky: Wow.
Eric Simons: And there’s many such cases of this. If you’re looking at greenfield projects, 100%. Today with this current state of frontier models, you could absolutely build production grade software. You’re not going to get a zero shot, but you’re going to spend a couple of days, weeks, whatever. But the cost reduction there, 30 grand versus $300. It’s 99% cheaper. Six months versus three weeks. I mean, it’s like order of magnitude sort of fashioned delivery on the thing. And those numbers have helped, for the people that we talk to, that are building these full stack apps. People, they go to Upwork, work, they get a quote for five grand. They have it within 50 bucks. It’s just nuts, what you’re able to do with this thing.
And so I think on the flip side, a lot of the existing companies, there are very legitimate use cases where things are greenfield, spun up. A good example is public websites. Marketing pages, landing pages, whatever have you. Folks are adopting Bolt to just power those instead of using Webflow, for example. Because it’s like, this is simpler to use than Webflow. And it integrates with the existing design system of the company, and et cetera. And the marketers can update it without knowing how to code, whatever. But then for product development teams, this is most commonly for, again, existing software businesses. They’re using this to just accelerate the product development process, and in a way where it’s not just like a Greenfield wholesale, “Hey, we’re building the entire thing in Bolt,” or whatever.
Lenny Rachitsky: Can Bolt integrate with your existing code base, or not yet?
Eric Simons: So yeah, we can actually open up repos in Bolt. You can go and use Bolt on your code base. It kind of depends on your setup. And we do have companies, again, that have marketing sites they’re using this on, or their admin panel or whatever. And I think it’s going to be a use case that we see a lot more people orienting towards. These LLMs are not great, depending on how big your application is, though. These things are not quite there, where if you have something that’s a thousand files or something, or more, where you’re going to be able to have a really reliable, super reliable experience per se. Within a year, we’ll chat a year from now, I suspect the answer is going to be different. So it kind of depends on the size of the app, the scale of the app, and if it’s too big, you’re looking at the prototyping, just pure acceleration of product development. And if it’s not, then you can just do it entirely from Bolt.
Lenny Rachitsky: So this is useful. So what would you say are the major limitations of Bolt today, where people should just know, “Okay, it’s not going to get you here yet. Maybe in the future it will.” So it sounds like, if you have a really large existing code base, probably not the best tool yet. What else should people know?
Eric Simons: I would say that’s probably the main one, because I think if you have a large existing code base, you’re going to need something like Cursor. And you’re going to need to be a developer, meaningfully, to be editing that stuff. I think outside of that, there’s a, just like using any other productivity tool, like Photoshop or Figma, or like a DSLR or whatever. There’s some level of education, and using the tool, and learning how to use it, that’s required to really unlock a lot of the maximum capabilities of the thing.
And the people that we see that are most successful with Bolt, outside of developers, the people we see that are most successful are people that are amazing PMs, for example. Because these are people that understand enough about how the technology works, typically, and their job is to direct developers on how to go and improve the product. And go and look into how to actually spec this thing out in a way that’s executable, without lossiness in the communication. And when you think about, “Okay, how would you best interact with an AI developer agent?” It’s basically that. You really want to be good at defining scope, and helping it go and debug various things, or whatever have you. There’s a huge overlap of the skill set of being a rock star PM, and being really good at using, frankly, any of these text to apps or Cogen tools.
Lenny Rachitsky: I love that you made that point. That’s exactly the point I’ve been trying to make, I have a newsletter post about this. Because when all these tools came out, there was so many people saying, “Okay, PMs are dead. We don’t need them anymore. We can just build things so quickly and easily, what’s the point?” But I completely see the world the way you see it. The hard part now is, now it’s easy to build the thing. Now it’s, “What the hell should we build? Can we clearly articulate what it is we want to build?” And then, “Can we just have the taste to know, is this right, is this correct? Is this good? Is this going to solve the problem?” And then it’s like, grow it, which is something also PMs think about. So, I completely agree. Basically it feels like PMs are, and a lot of PMs listen to this, so they’ll love hearing this. To me, it feels like PMs are the best positioned role to thrive in this world.
Eric Simons: Zero question. I mean that was, as Bolt was growing and we were like… Because we were a developer product before this, and so we expected the audience to be 100% developers that were using this. And we just kept seeing more and more and more people that were not developers using it, to the point where it’s like, 67% of our users are not developers, at this point. And when I started talking to these folks, at first I was just weird, or whatever. It was like, “Well, what’s going on here?” But then it just kind of clicked as like, “Oh, well, this is going to change everything. The entire software world order is going to get rewritten, here.”
Because the way that companies are organized to build software today, totally going to change. The idea that again, PMs are the people that really understand, to the pixel level, what matters into making a great product experience. And often they’re having… And listen, I’m a developer, myself. They have to go and harangue the developers to get things to be how they really ought to be, to the smallest levels. And now, how this is going to work, if you fast-forward one, two, five years, whatever. PMs, they’re going to be “writing code”, quote, unquote, instead of just writing a JIRA ticket and waiting for a developer to do it. The developers are going to be able to work on intellectually challenging tasks that LLMs are not well suited for, and still being augmented by LLMs to do it. But PMs are going to be able to go in and just make the changes themselves.
And what blew my mind is, it’s not priced in, to any of these companies out there. And it’s not reflected in the org charts of all the software companies in the world right now. That is going to completely change. The winners, at least, their org charts are going to completely change, and how they approach building products and shipping products. Completely. And it’s starting, this is the beginning.
Lenny Rachitsky: I want to follow that thread, but first of all, I want to also add, and correct me if you disagree with this. I think when we talk about PMs, that also applies to founders, like product thinking founders-
Eric Simons: One hundred percent.
Lenny Rachitsky: … very similar. And then, I think it’s also important to note, if you also have engineering skills and design skills, you will be at an advantage. That only helps you. But if you’re looking at this triangle of the triad of product, engineering, PM, it feels like the PME skills are the ones that will be most important and valuable. Although it’d be great if you can also be in code, and if you could also design really well.
Eric Simons: Absolutely. And to me, it’s the most exciting mix. I think PMs, designers and entrepreneurs that are non-technical, that’s the most exciting thing to me, just because it’s a brand new market that’s being unlocked here. For the first time ever, these folks can directly code and build the product, themselves. Their vision, directly into the software itself. That’s going to change everything. That is changing everything.
Lenny Rachitsky: So when you talk about how org charts are going to change, what are you imagining there? Is it just fewer engineers, mostly, or what do the future org charts look like?
Eric Simons: Good question. And I bet you there’s going to be some Gartner analysis someday, years from now or whatever that’s like, “Here’s how the best,” some term is applied to how the best companies are organizing. But yeah, I think that we’re going to see developers probably being pulled off of a lot of the, generally speaking, pulled off of a lot of user interface type work. I would imagine. Except for the most complicated of those things. And you’re going to see designers and PMs really, really leading the charge, and being responsible for crafting those experiences. And perhaps having a developer attached to be reviewing the code and making sure the guiding, the code that they’re writing, reviewing those pull requests and et cetera. And I think maybe even the engineers are… Like you pointed out, having engineering skills is not going to hurt you. It’s going to make you way more effective.
But I do think there’s going to be, the leverage that the front engineer is going to have is, it is now insane, it’s only going to get more so. And so I could see there just being fewer front engineers attached to, I’m seeing more product and design folks, with one or two engineers or something. And really having a larger matching of pods like that. Something like that strikes me as probably how this is going to start trending towards.
Lenny Rachitsky: This touches on, we had a researcher from OpenAI in the podcast. She actually started her career, she worked at Anthropic first, as a front-end engineer. And said that once she saw what Clyde could do, for front-end engineering, she’s like, “I need to move to a different function.” And so she moved into research, because she saw that role disappearing, potentially. And that’s exactly what you’re saying.
Eric Simons: Yep.
Lenny Rachitsky: So let me ask you this, I don’t know if you have a clear thesis on this yet, but say, you just had a kid. Say your kid is, in the future, starting school. Let’s say your kid was starting college soon. Do you have thoughts on just what skills slash areas you think they should go into, versus avoid, that maybe are popular now and are going to be less popular?
Eric Simons: Understanding how to leverage these AI tools is key. I wouldn’t necessarily, I think maybe getting a basic understanding of how programming works, et cetera, is great.
Lenny Rachitsky: It’s like technical foundations, just understanding how systems work, how coding works.
Eric Simons: Exactly. But it doesn’t have to be, because I think back to if Bolt existed, like Albert and I say this to each other all the time. Since the get-go, StackBlitz, we’ve been building the thing that we wish we had when we were 13. And heck, for everything we built since then. And especially with Bolt. I mean, I don’t know if I would’ve gone as deep as I did on learning how to code, and being an engineer, if that had been around then. The whole reason we got into it is we had ideas for products, and businesses, that we wanted to build. And coding was just a necessary requirement in order to do that.
And that said, I think people need to follow their intrinsic interests. If folks are really interested in really getting in the nitty-gritty of how computers work, and program leaders, or spark and compilers, or whatever, go for it. I think that stuff is still going to be relevant. I don’t know if we’re going to really have, we’ll see, but to the degree that there’s AGI where it’s like, we don’t have to think about anything ever again.
Lenny Rachitsky: Yeah, that’s always the answer here. Do we sell anything?
Eric Simons: Yeah, it’s like if we’re at that point, it’s kind of… I don’t know, I’m not sure. But I think from, at least what I feel like seems like the next at least five years of what we’re looking at, I think people are still going to, there’s still going to be places to specialize, and really go deep. But I think you want to go into it with the idea, not like, “I’m going to go and learn computer science because I’m going to get a job for sure out of it.”
I just think that’s generally not a good… This is like, my co-founder and I, we didn’t go to college. My co-founder dropped out of college after a semester or something, but I didn’t go, because I was like, “We’re coding,” we were doing contracting at the time, making money. And it was like, “This is a lot of,” you know, it would’ve been like a hundred grand of debt by the end of the thing, just for four years of in-state tuition, at U of I. 120 grand, I think, at that time.
And lo and behold, I mean there’s a huge issue with this. Where people are kind of… There was a prevailing thought by society that going to college in the early 2010s or late two 2000s, that you’re going to get a job on the other side. That’s going to be high paying. And that just has not been the case for a lot of people. And I think that’s just going to continue to be the case. But again, not to deter people from doing it, but you have to go into it being like, “I for sure, this is what I want, and I want to go and be the best that I can possibly be at this thing.” You know what I mean?
Lenny Rachitsky: I like the transfer to your kid is going to be like, “Don’t even go to college,” potentially.
Eric Simons: Only if they want to. I think at 18, it’s a huge ask. I mean, it’s a huge ask, not even at 18. It’s like at 17, because you could go apply for colleges. It’s just such a huge… Like, a six figure debt commitment to someone who’s making $0, or negative dollars, and that young. Unless you really have conviction, it costs nothing to go and explore and learn for free, online.
Lenny Rachitsky: I want to come back to the skills that you think are going to be most important, and let me try to mirror back a few things you said, that I very much agree with.
So it feels like if you want to be successful in the world where AI can build things for you, more and more, what I’m hearing is get good at figuring out what people need and want, what problems they need solved. Get good at articulating it really well to the AI tools. And there’s this talk, “You don’t need to be a great prompt engineer. You don’t need to work on prompting,” but it feels like it’s more and more important, because you tell it something and it goes off and builds a thing. If you’re clear about it, it’ll save you a lot of time.
So it’s be good at figuring out problems people need solved, figure out how to articulate that problem well, and ask for a clear solution. Figure out how to grow the thing, feels like that’s still going to be a need, because Bolt’s not going to go and find every… I could see that still running paid ads, and stuff like that. But it feels like that’s going to be ongoing need.
And then I feel like there’s this kind of unstuck step, like helping AI get unstuck, and it feels like that’s where maybe engineering skills will come in more and more. Thoughts on just that skill?
Eric Simons: Oh, totally. Yeah. So we actually, two weeks ago I think, we announced this program called Bolt Builders. And it’s basically the genius bar at the Apple Store, where as folks are building on Bolt that are not developers, they’ll run into some nook or cranny where the AI just cannot figure it out, or whatever. And I think that’s just going to continue to be the case for the time to come. That’s our position, and that’s why we spun up this program.
Lenny Rachitsky: And these are humans, that help you out.
Eric Simons: These are humans. And people that we’re certifying. And so, in Bolt in the coming weeks or whatever, there’s going to be a button where you can just say, “Hey, connect me with a certified expert.” And you can chat with them live, and they’ll help you get unstuck, and you pay I don’t know, 50 bucks an hour. Whatever it is. And then you get unjammed and you keep prompting. And again, I just think this stuff is, it just all seems like gravy to me. Engineers get to focus on difficult challenges, not like cookie cutter, “Let’s make another CRUD app,” stuff. They get to, debugging is challenging, and fun. And going and working on intellectually stimulating tasks, and all this stuff that’s just copy pasta, over and over, all this, error app. It’s just like, let the AI do all that kind of crap.
Lenny Rachitsky: This is a potentially new job, for now at least, just unstuck the AI. Which I think over time, it’ll get better and better, and we maybe won’t need these people. But I love that it’s now AI first, and person second. Versus person building the thing, and then AI. Like when Copilot launched, it was just like, “Cool, here’s a little suggestion for this function,” and now it’s flipped. “Here’s everything.” And then, “Oh, I don’t know what to do here. Help us here.” And then, it’s like a human suggestion. Isn’t that interesting? It’s like, human Copilot is flipping it.
Eric Simons: Totally. Yeah. That’s what’s wild is, I think Sonnet was really the first model that flipped the equation, because that was really us, and old Cursor, and all these other things. The rapid growth started the second Sonnet went online. We actually tried building Bolt almost exactly a year ago, with the frontier models at the time. Spent a week or two building it. It just didn’t work. The output, the code output was not reliable enough. It would constantly, it would be a broken app, or it would look ugly, or whatever. And then we got a sneak peek of the Sonnet stuff in May and we were like, “Oh. Okay, we should take that project back off the shelf and green line it, because this might be it.”
And lo and behold, that’s exactly what has happened. But yeah, that’s the big deal that is, kind of under the hood, this is… What’s going on here is, a very critical threshold has been passed with LLM’s ability to write production grade code and apps that actually look beautiful, and actually function well. It’s not perfect, but there’s kind of this zero to one moment that’s happened where it’s like, “Okay, so now, yeah. Now the AI is the first thing,” and then you’re kind of popping a developer in every now and then, versus the other way around.
Lenny Rachitsky: I did not know that. I didn’t realize that so much of this was unlocked with, like it’s sitting on top of Anthropix work, and specifically Sonnet. That was the first model, you’re saying, that could code well enough.
Eric Simons: Yeah, zero question.
Lenny Rachitsky: Wow.
Eric Simons: Zero question. Absolutely.
Lenny Rachitsky: That is fascinating. Just the amount of, I don’t know, revenue and business and ecommerce that that one model has unlocked, is insane. I did not realize that.
Eric Simons: It is. And in retrospect, I’d mentioned, we’d never done an AI product at StackBlitz, and it’s tempting. Like when of ChatGPT went online, everyone started adding AI to their products. We just didn’t see a clear place for a really added value. So I was not super bullish on, you know, a lot of people were like, “AGI is going to be here in 2023.” You know what I mean? There’s all this stuff that was being said, and I was like, “I just don’t know if I necessarily buy how fast people say that it’s going to move.” And to a certain degree, that was the correct view.
What I didn’t really think about though, is if AI, if LLMs are going to get better at a specific vertical, which are going to be the things that it would be. And if you look at law, for example, you want to make the best LLM for looking at case law. The problem with that stuff is, it’s not deterministic. The judge’s ruling is dependent on society’s view of things at the time, political stuff going on, the jury. There’s a ton of things, that it’s not deterministic. And so you can’t really create a lot of training data that’s going to be super reliable, and produce really good results. And you can’t just make up cases and say, “Theoretically, the judge would say this.” Because, I mean, it just doesn’t work.
Software is deterministic. When you write code and you hit run, it either runs or it doesn’t. And that’s the key insight Anthropic really had. They just went deep. And then, this is what they’re doing, is just reinforcement learning on basically permutating every type of app you could ever build, and just spinning up tens of thousands of cores or whatever to do that. Just building tons of training data, and doing reinforcement learning, and making their LLMs the best in the world at building beautiful, reliable applications. I’m extremely bullish. It makes technical sense why, of anything, LLMs are going to get insanely better at writing code than probably most other types of applications for LLMs. Simply because it’s something that can be extremely deterministic, and permutated thousands and thousands and thousands of times per second.
And so I think the broader trend here is… And Sonnet has woken everyone up. Google, and Open AI, all the… Everyone is now gunning for coding because, how big is the market opportunity to rewrite the software world order? It’s trillions of dollars, or something, right? The world runs on software. So I think that just in a macro, the highest macro view, and why we went and raised money for Bolt. This seems like an extremely clear shot, there’s this, you can kind of separate the hype of what people say and blah, blah, blah. When you break it down technically, this makes sense. It makes sense what’s going to happen here. And for us, we’re like, we are so happy to be well positioned to go and enable people to kind of ride this wave of the innovation that is here in LLMs, and is going to just keep coming, and therefore enable more people to build even crazier amazing software. So that’s our world view, at least, of what’s going on here at the macro.
Lenny Rachitsky: And when did Sonnet even come out? It’s been a while, right?
Eric Simons: I think they officially, I think it was in June, when they officially put it online.
Lenny Rachitsky: So since June, this is the worst it will ever be, the state of AI coding. And it’s already this good. And there hasn’t been anything, like they haven’t launched their new model, since last June. So this tells us just how quickly things are going to start moving once they launch their next model. And as you said, everyone’s gunning now for this, because they realized “We’re behind on the coding piece.” So, wow. This is going to get crazy.
Eric Simons: I agree. Yeah, it’s been, again, there was no blog post that laid all of this out for us. It’s just been kind of this-
Lenny Rachitsky: You just noticed the code was really good, basically.
Eric Simons: And from there, it’s just, we’ve been piecing together all this other stuff. So it’s been kind of the thrill of a murder mystery of, “What is going on here?”
Lenny Rachitsky: Oh, really.
Eric Simons: Yeah. You know what I mean?
Lenny Rachitsky: Just piecing together what you’ve seen on Anthropic releasing, is that what you mean? What have you been noticing, murder mystery-wise?
Eric Simons: Well, and then the impacts of Bolt, where we have people that are not technical using this. How are they using it? Why are they doing this? And then, all the stuff we’ve talked about in this podcast has been the result of nine months of just R&D and seeing the results of it, and then going, “What?” And then digging in, doing another thing and then going, “What,” again. And then it just keeps happening, because there’s no charted course for this.
Lenny Rachitsky: It’s like an anthropology, if that’s the right term, of just watching. It’s like an emergent discovery, it sounds like. Versus you had the strategy, “Here, we’re going to do this,” where a persona to launch this thing will happen?
Eric Simons: Yeah, a hundred percent. Yeah. I think that that’s the best way to put it. It’s the best way to put it. It’s exactly that. And so it’s very interesting to just-
Lenny Rachitsky: To watch, and be a part of it, I imagine.
Eric Simons: Yeah, yeah.
Lenny Rachitsky: And I think as people say, “Okay, these are just toys, they’re prototypes, it’s not going to work with your existing code, it’s not going to scale.” It’s important to note just what we talked about. This is a model from last June that this is possible on, and everybody’s working on the next cutting edge model that will make this even better, and that’s going to come real soon. Okay. Amazing.
I just have a few more questions to close out this conversation. One is just, what is coming next for Bolt? What are some of the cool new features that’ll be launching before this comes out? Maybe right after this comes out, maybe in mid-March?
Eric Simons: Yeah. Okay. So by the time, and I’m going to go back and tell our engineers, “I said this on this podcast-”
Lenny Rachitsky: “I’ve committed. Sorry, guys.”
Eric Simons: This is, I found actually being a leaky faucet on talking on podcasts and stuff, my engineer is like, “How could you tell them…” “You just have to ship it faster, now. You got to make it real,” right?
Lenny Rachitsky: Or we’ll get AI on it.
Eric Simons: No, no. Yeah. But I think for us it’s like, again, we’ve seen a lot of PMs, designers, entrepreneurs, et cetera, using Bolt. And so, we’re really looking at better fitting with the tools that folks are using to do those things today. Bolt’s not going to replace them, or something. And if you’re working at a company, how do you integrate this stuff with your existing business, or your existing product and code base? Because that’s the question we often get from people is, “How do I open my,” like we’re talking to one of the fan companies the other day, and like, “How do we open our production code base in this, that’s like 20 years old?” I’m like, “You don’t. None of this stuff. That’s not what you do. This is for rapid product development in your use case.”
So the features that we’re going to be shipping, I’m pretty stoked about this one, so we’ve been working on this for a while and we’ve partnered up with a company called Anima to do this. But basically, so on any Figma URL, when you’re looking at a design that you’ve made, if you just put bolt.new in front of that URL and hit enter, it’s going to suck that design into Bolt, and turn it into a full stack app or mobile app, just out of the box.
Lenny Rachitsky: That is genius.
Eric Simons: Yeah. It’s-
Lenny Rachitsky: Amazing.
Eric Simons: It’s going to be nuts. Yeah, I mean, it’s really fun to use. Because whether you’re a developer or designer, or whatever, going and taking that and turning it into an actual coded app. And the thing is, once it’s in Bolt, you can just keep prompting from there. You’re like, “Yeah, well, add another page here.” So you can have things that, where you want pixel perfect design, you can have it, and it’ll translate one to one. And it’s splitting out the assets. Anima’s been doing this since 2017, Figma to code. They’ve got the best agent in the world for, they’re the number one Figma plug-in, or whatever. And so in Bolt, it’s going to just work. It’s just deeply integrated
Lenny Rachitsky: So it’s bolt.new slash, the Figma URL, to the design.
Eric Simons: Yep. That’s all you got to do.
Lenny Rachitsky: Amazing.
Eric Simons: And in Bolt, we’re going to also have a little Figma icon in the chat thing. So if you go to Bolt itself, you can click it and then paste the URL, or whatever. But yeah, but it’s like that. It’s from Figma to full stack app in a click. Literally. That’s crazy. So that one’s pretty cool.
And then the other one that we’re working on is an integration with Slack, because often when you’re talking about Figmas, like Figma links or whatever, at a company, you’re in Slack or whatever. So you’re having conversations about, “Hey, we should really add a page to this that does da-da-da.” And so we’re actually creating a Bolt Slack bot whose job is to basically act like a developer on your team. And so you can be, in a thread, you can be like, “Hey, I think we should a homepage.” “Yeah, okay. @Bolt, can you whip this up real quick?” And it’ll go and just suck down the entire conversation history so far in your Slack, on the thread or whatever.
You’ll be like, “Okay, cool. So you kind of want me to take this Figma URL, which I can convert automatically,” thanks to the future I just mentioned right before this. “I can just go and convert that thing out of the box, and then you want me to add a page to it, and then do this thing. Got it. I’m going to go do that. Oh, here’s the URL where you can open it up and keep prompting.” You know what I mean? So it’s like having a developer or somebody to kind of kick this thing out. And you’re just like, “Go make this thing real quick.”
So those two things, I think it’s, again. You start think about how our company’s going to change how they’re currently doing product development. And even just like, “Hey, we need to spin up a marketing site. Here’s the thing.” We’re like, “Can you do that, Bolt?” “Yep. Let me go do that.” You know what I mean? And that’s why I’m kind of excited about those two, in particular, because I think it’s going to be well received, and folks are going to be stoked about it. I hope. Knock on wood.
Lenny Rachitsky: Those are awesome features. I love the Slack piece, because when I think about agents, there’s always talk about agents. To me the simplest way to understand it is just a Slack bot, just that AI can talk to you like they’re a person in Slack. And I love that’s exactly what you’re doing. And this is this gigantic feature of just like, “Hey, you just have this engineer now that can go build stuff for you.”
Let me actually ask you a question along these lines, that I was meaning to ask, but I forgot. Is just, your engineering team, what are they using to build Bolt? I imagine it’s a lot of Cursor. How much is Bolt, at this point, involved in building Bolt? And is there any other tools that they found useful, find useful, that are worth highlighting?
Eric Simons: Yeah, good question. Yeah, we definitely use Cursor. Our folks use Cursor a lot. We use Bolt a lot for the product development process, like a ton, we’re using it. And we’re doing basically the flow that I described, where if things that need to be Pixel perfect, we’re going to Figma for. And often we’re taking that and we’re pulling that into Bolt, because we’ve got access to the integration today. So pull that into Bolt and we’re saying, “Hey, go add these things or whatever.” Or just saying, “Hey, here’s a screenshot of our UI, go do da-da-da.”
Other AI tools that the developers are using. I think those are the primary ones. I mean, I think we’ve got a subscription to Claude, and ChatGPT, and things like that. But I think for development, Cursor is the main thing,
Lenny Rachitsky: Yeah, it’s cool how few tools, like there’s so many AI tools, and it’s interesting how few people actually end up using. It’s like Cursor, Claude, ChatGPT, and then maybe another tool. Like Bolt.
Eric Simons: Yeah. Totally.
Lenny Rachitsky: Okay. Final-ish question. Say somebody is opening up Bolt for the first time. What’s something that, imagine you could sit next to every new user that’s just trying Bolt for the first time, and you could whisper a tip in their ear to be successful with Bolt. What would that tip be?
Eric Simons: And this is like, because we have a lot of different types of users. I imagine you’re talking about PMs or designers, and that sort of-
Lenny Rachitsky: Let’s do PMs. That’s a good one. That’s a lot of the audience here.
Eric Simons: I would say, talk to this thing like you do a Linear ticket, or a JIRA ticket. That would be my advice. And talk to this like you would, like you’re talking to one of the developers on your team. And what that means is, be specific on things that matter. And on things where, also, you can let it be creative. You can go to and just say, “Hey, make it prettier.” And it does a good job, it actually does a really good job, when you give it just, vibes. So anyway, I think for PMs it’s like, you have the skillset. You know how to do this. This is, just think of this as your coworker, your developer coworker.
Lenny Rachitsky: I love that. Because these tools are so easy, you just go in and it’s like tell it a thing, and then, cool. You have a website. It’s coded, boom, done. And what I’m hearing here is, take a little time to craft your ask. It may be tempting to just start, “Cool. Build me a serum,” and then you’re stuck with that first version. And then you’re like, “Oh, well, okay. I didn’t mean that.” So it sounds like your advice is take some time to craft the ask, and be clear about what you want.
Eric Simons: Yeah, totally. Especially if you have a clear vision of what you’re trying to build. And something reasonably sophisticated. And what I recommend everyone to do, if it’s your first time trying Bolt and you’re like, “What should I have it do, and I don’t have an idea”? Tell it to build you a personal website. There’s something like magic. You take your LinkedIn copy, and paste your LinkedIn bio and work experience, just like select text, copy, paste it. “I need a website, my name is so-and-so, here’s my LinkedIn history. My favorite color is blue, and I like dogs.” And then hit paste, right?
Lenny Rachitsky: “Make it prettier.”
Eric Simons: [inaudible 01:22:24] actor. Yeah. And then, you know what I mean? And then you can hit deploy. And if you don’t have a .com yet, now you can. Right? I mean, now you have a real, personalized service. I think there’s kind of a moment around that where it’s like, “Oh, okay, wow. This,” it’s zero shot, zero shot 99.999% of the time. You’re getting a beautiful personal website that you didn’t have before, that would’ve taken you an hour, on Wix. If not more. And that gives you the taste of, “Oh, okay, cool. So if I really take the time to think this through, and make a PRD, and then put that in piece by piece into this thing, the sky’s the limit.”
Lenny Rachitsky: Eric, this has been just insane on so many levels. I have so much to process, I think a lot of listeners do too. Maybe as an actual final question, I saw this story about how when you were starting StackBlitz, or maybe even before StackBlitz, you squatted in the AOL office. Because you had some badge that still worked. Maybe just tell that story.
Eric Simons: Yeah, that was the thing I was most known for. That happened like 2012, and I was 19 years old, so it’s been a very long time. I think I’m 33, now or something.
Lenny Rachitsky: The statutes of limitations are expended.
Eric Simons: Yeah, yeah, exactly. But yeah, for a while I was like, “I have to do something more notable than living at AOL. This can’t be what I’m known for.” But now people are like, “Oh, wait. You’re the StackBlitz guy, and you’re the AOL guy?”
So when I first came out to Silicon Valley, we got into a, I was building a K12 education startup at the time, and this is back during the days where Y Combinator, they only gave you like 20 grand, and so there was this offshoot of Y Combinator called Imagine K12, it was Jeff Ralston who, I think he was CEO of YC a couple of years back. And then Imagine K12 merged into Y Combinator a couple of years after Rob went through it, but anyways, they had an office space at AOL. At the time, AOL was trying to reinvigorate the company and they were like, “Hey, we should get young startup blood in here, so let’s rent out office space to teenagers,” basically.
So I was there, and we ended up running out of money. 20 grand doesn’t go very far in the Valley, so three or four months in, we were like, “Oh God, what do we do?” And I was going to the AOL office multiple times a week, because we had access cards to get in to get to the investors’ offices. And I realized, I was like, “You know, they have couches here, and they have food. There’s ramen that you can microwave, and there’s a gym where there’s a shower, and even you can do laundry.” And then, so I was like, “I don’t know, maybe while I figure this out, I’ll just live out of here.” And so that’s what I ended up doing for, I think, four or five months. I was living out of this headquarters over on Page Mill in El Camino, in Palo Alto.
And then, I got away with it for a while just because the guards, the security guards, they worked 12-hour shifts. And so the guys that, when I was there at night… And I was coding, all day every day, basically. So the guys at night just were like, “Dang, this guy works really hard.” And then in the morning they’d be like, “Wow, this guy is working really hard.” And I became friends with some of them, and then eventually, I think there were also a whole bunch of Stanford students that I think they put bunks in one of the aisles. It was just started getting out control, so I think they started cracking down. And then one morning, at like 4:00 in the morning, a guard came in and threw me out.
I’m from Chicago. I don’t know anyone. At that point I’m like, “I know no one in the Bay Area.” So I went to a Starbucks, which was not open. I slept on the table outside of the thing. And I think I hit up one of the other entrepreneurs that was in the program. I was like, “Do you have a couch? I think I kind of need it, at this point.” Yeah, the press got wind of it, and it was this worldwide story. But I lived on a dollar a day. That was the crazy thing. My burn rate was a dollar a day, at that time.
Lenny Rachitsky: What did you use that dollar for?
Eric Simons: This is back when McDonald’s had the Dollar Menu. Literally. So it was like, I occasionally would go and get a cheeseburger or whatever. Yeah, it was ultimate scrappiness.
Lenny Rachitsky: You’re technically homeless. From homeless, to one of the fastest growing startups in history. Eric, what a journey. This is such an interesting point in time of your life, and of just tech. No matter what happens, I’m sure you’ll be extremely successful, but it’s such an interesting just point in that journey. And I’m thankful that you made time to share it with us.
Eric Simons: It’s always good to just have the perspective of, you should start companies to keep the mindset that you’re doing it to have fun. So, stoked to see where this goes, one way or the other. It’s going to be interesting.
Lenny Rachitsky: Eric, final questions. Where can folks find online if they want to reach out, maybe follow up on some stuff you shared, and how can listeners be useful to you?
Eric Simons: Yeah, totally. I mean, yeah, so bolt.new is the website, over on Twitter, I’m @ericsimons40 on Twitter, and I think our Twitter account is @boltdotnew, not with a period. It’s like, B-O-L-T-D-O-T-N-E-W. And yeah, I’m curious to hear what folks think. I mean, this is, again, we are learning so much from the people that are coming and trying this thing out and giving their feedback. And within the first meeting of it going online, we were not the experts on how use the tool anymore, and it’s been that way ever since. And so, I love hearing from folks on what they want to see next, and how this is helping them. And where they run into problems, like where we need to go and fix things. So my email address is Eric@stackblitz.com. That’s Eric with a C. So I’d love to hear from anyone, whether it’s a DM on Twitter, or an email.
Lenny Rachitsky: Amazing. Eric, thank you so much for being here.
Eric Simons: Awesome. Thank you so much for having me. This is a blast.
Lenny Rachitsky: Bye, everyone.
Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review, as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at lennyspodcast.com.
See you in the next episode.
Glossary
| English | 中文 |
|---|---|
| AJAX | AJAX(Web 异步通信技术) |
| Albert | Albert(Eric Simons 的联合创始人) |
| ARR (Annual Recurring Revenue) | ARR(年度经常性收入) |
| Bolt | Bolt(产品名称,保留原文) |
| Canva | Canva(产品名称) |
| Cloud 9 | Cloud 9(早期云端 IDE 产品) |
| Copilot | Copilot(产品名称) |
| CRM | CRM(客户关系管理系统) |
| CSV | CSV(逗号分隔值文件格式) |
| Dalton | Dalton(YC 合伙人) |
| Dario | Dario(Anthropic CEO Dario Amodei) |
| DAU | DAU(日活跃用户数) |
| DSLR | DSLR(数码单反相机) |
| Dylan Field | Dylan Field(人名,保留原文) |
| Eric Simons | Eric Simons(人名,保留原文) |
| ERP | ERP(企业资源计划系统) |
| Evan Wallace | Evan Wallace(人名,保留原文) |
| Expo | Expo(公司名称,保留原文) |
| Figma | Figma(产品名称,保留原文) |
| Gartner | Gartner(IT 研究与咨询公司) |
| Google Docs | Google Docs(产品名称) |
| GPU | GPU(图形处理器) |
| JIRA | JIRA(项目管理工具) |
| Lenny Rachitsky | Lenny Rachitsky(人名,保留原文) |
| Linear | Linear(项目管理工具,保留原文) |
| Netflix | Netflix(公司名称) |
| Netlify | Netlify(服务名称,保留原文) |
| Node.js | Node.js(JavaScript 运行时环境) |
| Notion | Notion(协作工具,保留原文) |
| NPM | NPM(JavaScript 包管理器) |
| OneSchema | OneSchema(产品名称,保留原文) |
| Photoshop | Photoshop(图像编辑软件) |
| PRD | PRD(产品需求文档,Product Requirements Document) |
| React Native | React Native(技术名称,保留原文) |
| Service Worker | Service Worker(浏览器 API) |
| SFTP | SFTP(安全文件传输协议) |
| Squarespace | Squarespace(产品名称,保留原文) |
| StackBlitz | StackBlitz(公司名称,保留原文) |
| Stripe | Stripe(在线支付服务) |
| TechCrunch | TechCrunch(科技媒体) |
| Upwork | Upwork(自由职业者平台) |
| Visual Studio | Visual Studio(产品名称) |
| WebAssembly | WebAssembly(技术名称,保留原文) |
| WebContainer | WebContainer(技术名称,保留原文) |
| Webflow | Webflow(网页设计工具) |
| WebGL | WebGL(浏览器图形渲染技术) |
| Wix | Wix(产品名称,保留原文) |
| Xcode | Xcode(产品名称) |
| YC | YC(Y Combinator,创业加速器) |
Reformatted by reformat_english.py
深入 Bolt:从濒临死亡到历史上增长最快的产品之一 | Eric Simons
深入 Bolt:从濒临死亡到历史上增长最快的产品之一 | Eric Simons
对话记录
Lenny Rachitsky: 你们的增长速度太离谱了。你们身处这样一批公司之中——增长速度在创业史上前所未见。
Eric Simons: 推出 Bolt 的时候,公司几乎要倒闭了。结果发生的事情是,前两个月 ARR(年度经常性收入)从零涨到了 2000 万美元。我们现在已经突破了 3000 万美元 ARR,按照目前的增速,我们今年的目标是达到 1 亿美元 ARR。
Lenny Rachitsky: 这一切简直就是不停歇的疯狂剧情。这怎么可能?是什么让你们以这么小的团队,实现了这么快、这么大规模的增长?
Eric Simons: 最重要的是人。很难找到一家创业公司,核心的五、六、七个人能在一起待五年以上。
Lenny Rachitsky: 你们基本上是先做技术,再去找要解决的问题,而这通常是人们建议不要做的事情。
Eric Simons: 我觉得这就是做创业者最难的地方。有些时候你必须做出判断,而这些判断不会是共识。你得对自己的信念有信心,相信怎样才能把手中的牌打好。
Lenny Rachitsky: 很多人看到这些数据,却有时忽略了在此之前还有多年、多年的努力。
Eric Simons: 就好像是——Bolt 是一个”一夜成名”,花了七年才实现。
嘉宾介绍
Lenny Rachitsky: 今天的嘉宾是 Eric Simons。Eric 是 StackBlitz 的联合创始人兼 CEO,这家公司做了一款叫 Bolt 的产品,目前正与 Cursor 并驾齐驱,角逐历史上增长最快的产品之名。他们目前是最受欢迎的网页端 AI 编程应用,注册用户超过 300 万。去年十月上线两个月后,他们的 ARR 就达到了 2000 万美元。截至本期录制时,他们正在逼近 4000 万美元 ARR。Bolt 的故事非常传奇——他们实际上七年前就创办了这家公司,当时快要弹尽粮绝、准备关门了。但他们意识到,过去七年一直在打造的 WebContainer 技术,恰好非常适合在浏览器中构建 AI 产品。于是他们发了一条推文就上线了产品,用 Eric 的话说,这是一个”花了七年才实现的一夜成名”。如果你想深入了解 AI 编程应用的前沿,以及 AI 与产品构建的未来走向,这期节目不容错过。
好了,有请 Eric Simons。
Bolt 是什么
Lenny Rachitsky: Eric,非常感谢你来做客,欢迎来到播客。
Eric Simons: 谢谢邀请。嗯,我很兴奋能来这里。
Lenny Rachitsky: 对于那些不太熟悉 Bolt 的听众,Bolt 是什么?
Eric Simons: 其实很简单。你打开它,看到一个文本框,然后告诉它你想构建什么。不管是网页应用还是移动应用都可以。所以它算是过去几个月里变得相当流行的那种”文本生成应用”的工具之一。而且它不只是搭建一个静态网站之类的,你实际上可以用它构建全栈的、真正的软件,包含数据库、托管等等,全部通过提示词完成。而且速度快得惊人——不是说你花好几个小时、好几天去拼凑,一分钟内你就能看到结果。
增长数据
Lenny Rachitsky: 让我们来分享一些数字,看看你们所构建的业务的规模。你们的增长速度太离谱了。你们身处这样一批公司之中——增长速度在创业史上前所未见,而你们就在这批公司的最前沿。来聊聊上线时的数据,以及现在的状况吧。
Eric Simons: 好的,上线的时候,我们推出 Bolt 时,我们公司 StackBlitz 几乎要倒闭了。我们已经存在了七年,一直在做基于网页的开发环境相关的东西。所以当我们推出这个产品时,我们想着,“如果这能在接下来几个月带来 10 万美元 ARR,那就太棒了。“结果发生的是,前两个月我们从零涨到了 2000 万美元 ARR。我想现在大概是第四个月,或者四个半月左右,我们已经突破了 3000 万美元 ARR,而且马上就要突破 4000 万了。等这期节目播出的时候,看起来我们会达到 4000 万美元 ARR。所以,收入增长的规模简直疯了。
用户增长与团队规模
Eric Simons: 当然,这也伴随着疯狂的用户增长。过去几个月我们就新增了 300 万注册用户,月活跃用户现在大约在每月 100 万左右。所以简直就是……我做了 15 年创业,从来没见过这样的事情。和我聊过的每个人,我们的投资人等等,都很难找到与这里发生的事情相类似的先例。而且这真的很不寻常,因为我们公司六个月前根本没在做 AI 相关的东西。我们没有 AI 产品,然后几乎是从濒临倒闭到一跃成为全球最大的、完全基于网页的 AI 代码生成应用,无论是按流量还是收入来看。我觉得在代码生成领域唯一排在我们前面的创业公司就是 Cursor,至少在收入方面是这样。
所以总之,这一路简直疯狂。而我们团队只有 15 到 20 人,我们马上就要突破 10 万付费客户了,而客服团队只有三个人。所以我们正在尽全力快速扩张。需求的规模简直令人难以置信,我们不得不以最快的速度调整一切来尽可能跟上需求。
Lenny Rachitsky: “难以置信”这个词用来形容你刚才分享的一切再合适不过了。100 万月活跃用户,4000 万美元的年度经常性收入,而产品上线才五个月。是这样吗?
Eric Simons: 对,个位数——个位数周。这是我们目前看到的增长速度。是的。
Lenny Rachitsky: 我觉得,你们是不是有史以来增长最快的创业公司?
Eric Simons: 嗯,我觉得这可能取决于你怎么定义这个数字。因为我们的目标就是做好产品,把技术的可能性推到极限。我觉得只要我们把工作做好,疯狂的事情就可能发生。但按照目前的增速来看,我们的 Q1 实际收入将超过预测,而我们全年的目标是达到 1 亿美元 ARR。如果达到的话,这将与 Cursor 持平,或者超过他们。
增长的可持续性
而且我认为未来还会有更多类似的情况。我只是觉得……确实有很多人对此持怀疑态度,他们会说,“好吧,还行吧。“这还是在我们第一个月就达到 400 万、500 万美元 ARR 的时候。当时我跟人聊天,他们会说,“好吧,是的,但这可能归零。“然后第二个月涨到了 2000 万,“啊,这也可能归零。“但现在我们已经逼近 4000 万了。
所以从我的角度来看,我一开始也非常怀疑。我从没见过什么东西这样增长,对吧?所以有一个月的时间里,我每天早上醒来都在等那一天——“好吧,结束了。“你明白我的意思吗?这件疯狂的事情发生了,然后它就不会再继续了。但那个数据始终没有出现。你在 Cursor 身上也看到了同样的情况,在很多其他 AI 创业公司身上也看到了。这个价值主张是真实的。自由市场里充满理性的参与者。人们涌向这些工具,是因为它们确实在解决问题,让人们能以更低的成本做到更多的事情。这也是为什么我说,我认为我们会看到更多这样的情况,无论是在编程领域还是其他垂直领域,或者其他什么地方。从某种意义上说,随着 AI 持续进步,这几乎可能成为一种新常态。不过,先不说这些。
Bolt 产品演示
Lenny Rachitsky: 我们来看看 Bolt 的演示吧,让大家亲眼看看它实际运行的样子。在你演示的过程中,如果可以的话,也指出一下与这个领域其他产品的区别,比如 Lovable、VZERO、Replit 这些大家可能听说过的产品,那会很有帮助。
Eric Simons: 好的,酷。这就是 Bolt,你直接访问 bolt.new 就行。我觉得 Bolt 真正有意思的几个点。第一,它极其简单。无论你登录还是没有登录,界面都一样,非常简单,就是一个文本框。而 Bolt 和其他产品之间最大的区别,其实很微妙。它不一定是你在界面上能直接看到的东西,而是它有多快,以及有多可靠。
这是因为我们处理计算的方式不同,因为不管是 Bolt 还是其他产品,当你输入内容时,它都需要启动一个开发环境来实际构建那个应用。所以需要在某个地方有一个操作系统来运行它。其他所有人都是把这些东西跑在云服务器上,这些服务器可能需要几分钟才能启动,而且经常会出问题,然后你可能真的会卡住,不得不联系客服来解锁。
而 Bolt 呢,过去七年我们公司一直在做的是构建一个在你浏览器本地运行的操作系统,使用的是你的 CPU。所以我们有一个非常宽松的免费层级,而且速度快得离谱,稳定性也极高。
所以如果我想要——就拿这个做一个快速演示——说”做一个 Spotify 的克隆版”,然后直接回车。这个东西已经开始工作了,右边这里已经是一个完整的开发环境。这是一个真正的操作系统,在我的浏览器里运行。我可以在上面运行命令等等。你在这里看到的,下面这个终端以及支撑它的底层,这才是我们真正花了五六年、甚至七年时间去构建并做到如此可靠的东西。如果没有 WebContainer 这项技术,就不会有 Bolt——它让我们能够在浏览器中运行一个操作系统。
这里发生的事情是,Bolt 的 AI 代理与这个操作系统之间是双向通信的。它在写代码,在运行开发服务器,接下来会把这个应用启动起来。你可以看到这有多快,大约 60 秒内,我说了一句”给我做一个 Spotify 克隆版”,现在就已经有了。而且看起来还挺不错的。
Lenny Rachitsky: 看起来真的很不错。
Eric Simons: 这也是 Bolt 的另一个亮点。我们为操作系统端开发的这项技术,过去五年多一直在做这个的团队成员,在此之前他们其实是在做机器学习和 AI 相关的工作。所以当需要为 Bolt 编写代理的时候,我们在内部拥有极其丰富的专业知识,知道如何真正把这两套不同的技术融合在一起,从而实现这种非常可靠的体验,产出非常漂亮、非常实用的东西。这就是 Bolt 体验真正酷的地方。
另一件事是,很多这类产品,你可以做出东西来,但通常你会想实际拥有一个可以分享的 URL。可能还想绑定一个域名什么的。而 Bolt,我们内置了与生产级托管服务提供商的集成,比如 Netlify,以及用于数据库的 Supabase。
一键部署到生产环境
Eric Simons: 所以如果我直接点击这里的部署按钮,它实际上会为我们做的这个项目运行一次生产构建。而且再说一次,这一切完全在我的浏览器里完成,所以对我们来说没有任何成本。所以你可以免费做这件事,而且它已经把这个项目部署到了一个真实的 URL 上,托管在 Netlify 上。这是实时上线的,我可以分享给任何人。如果我想购买 spotifyclone.com 这个域名,然后指向这个项目,我可以点击这里的链接。那会跳转到 Netlify,我可以把这个项目绑定到我的账户,购买域名,指向这个项目。然后从那以后,每当我在 Bolt 里让它对这个应用做修改并点击部署,那些改动就会直接上线到我公开的网站上。
这是有史以来构建 Web 应用最简单的方式。这也是我在产品上线几周后的一个关键领悟。我看到人们用它来做各种个人用途的场景,比如医疗捐赠网站、婚礼网站之类的。我当时就想,“这些人不知道有 Wix 或者 Squarespace 吗?我该不该告诉他们?“然后我突然明白了。那些东西实在太复杂了。我不知道你有没有看过它们的界面,但真的复杂得离谱,而且那还只是建一个静态网站。你根本不可能用它们来构建一个真正有功能的应用。而在 Bolt 里,如果我们再坐在这里花三十分钟,我们就能加上流媒体播放功能。你可以创建不同 MP3 文件的播放列表什么的。你只需要不断给它提示,它就会持续添加功能。
以上就是 Bolt 核心体验中一些比较酷的部分。我可以给你展示一个我们刚上线的新功能,如果你感兴趣的话。
Lenny Rachitsky: 来吧。
原生移动应用开发
Eric Simons: 刚才展示的是 Web 应用,对吧?Web 应用很棒,但很多时候你会想要一个原生应用。构建 Web 应用本身就够难了,构建原生应用就更难了,尤其是你还要把它放到应用商店里。所以我们和一家叫 Expo 的公司合作了,他们整个业务就是做 React Native 工具链和生态系统,让构建漂亮的应用并真正上架到应用商店变得非常简单。
就在这里,我稍微放大一下,我们有这个小小的”用 Expo 构建移动应用”的选项。点击之后,它会引导你如何通过提示来创建移动应用。那我们来做一个 Spotify 克隆,一个真正的原生移动应用。我们输入”做一个 Spotify 克隆”,然后回车。这个工具接下来会做的事情是,同样在这里启动一个操作系统,启动 Expo 工具链,然后真正为我们构建一个移动应用。
这个功能酷的地方在于,我们可以直接在浏览器里预览它,但等这个东西构建完成并启动之后,它会显示一个二维码,我们可以扫码,然后实时地获得这个原生应用的测试版,可以在我们的手机上试用,而且随着我们不断发出提示,你会看到它在做修改。这可以说是第一次,你不需要具备技术能力就能构建生产级的 Web、全栈 Web 和移动应用。到现在为止,我做的所有事情都不需要任何开发者知识。
我觉得这也是很多人对此感到兴奋的原因。我们的用户中大多数并不是开发者,而是 PM、设计师、创业者。因为这些人一直都擅长构建产品,但以前,他们把想法变成代码软件的唯一途径就是通过开发者的双手。而现在,他们可以通过自己的提示来实现。
你可以看到这里,我们有了这个二维码。我去扫一下。
Lenny Rachitsky: 我也来扫一下。
Eric Simons: 好的。
Lenny Rachitsky: 顺便说一下,我很欣赏你刚好在它完成之前有足够多的话可以说。太专业了。
Eric Simons: 正如我计划的,对吧?我的屏幕上它正在启动,正在打包这个应用的 JavaScript。这还是测试版,顺便说一下,我们上周刚发布的。你可以在我屏幕上看到,我实际上已经有了这个 Spotify 风格的应用了。
Lenny Rachitsky: 哇。
Eric Simons: 对吧。
Lenny Rachitsky: 看起来和 Spotify 一模一样。
Eric Simons: 看起来和 Spotify 一模一样,对吧?
Lenny Rachitsky: 很好。
Eric Simons: 是的。
Lenny Rachitsky: 我们现在要被告了,所以……你做得太好了。不开玩笑,真的很厉害。
Eric Simons: 是的。很酷,对吧?而且酷的地方是,随着你在设备上不断提示,它会不断重新加载。你不需要关闭应用,就能看到新功能被添加上去。所以在你我现在的这个场景中,就好像如果我们在一起开发一个应用,我们可以在地球的两端,你不只是看到这个东西的截图,而是真正可以触摸它、感受它,并且实际测试它的功能。
所以很多产品团队……这正在改变人们做产品开发的方式。这样做比设计一堆 Figma 画板还要快,真的。
零到四千万 ARR 背后的七年积累
Lenny Rachitsky: 我们待会儿会花很多时间聊这个。好的,这太不可思议了,到目前为止这整期节目就是你一次又一次地让我震撼,我相信听众们也是一样。有时候我甚至不知道该往哪个方向聊了。
你刚才提到了一个非常关键的点,就是在推出 Bolt 之前你在这个方向上做了七年。很多人看到这些数据,五个月左右从零做到四千万 ARR,但有时候他们没看到在此之前还有多年的积累。你们之所以如此成功,原因就是之前所有那些工作,就是你们开发的 WebContainer 技术。关于这段经历中 WebContainer 这个部分,你觉得有什么值得分享的吗?我知道我们后面会聊 Bolt 的完整起源故事,但单就 WebContainer 这个部分,我觉得这是个大事件。
Eric Simons: 百分之百是这样的。而且我觉得令人意外的是,这仍然是我们公司比较反共识的一个观点。因为这么多年来,当我们最初……WebContainer 就是我们把整个公司押上去的赌注。说清楚一点,StackBlitz 从一开始就是一个基于浏览器的深度技术押注——“我们能不能做一个基于 WebAssembly 的操作系统,能在浏览器里大概一百毫秒内启动,并且运行完整的开发工具链?“这就是核心问题。
我们之所以有这个想法,以及意识到这可能行得通,是因为当年我和我的联合创始人刚到硅谷的时候——他和我从小在芝加哥就在同一条街上长大,我们十三岁就一起写代码,从那以后一直在做东西。2012 年我们来到硅谷,有幸在早期遇到了正在做 Figma 的 Dylan Field 和 Evan Wallace。我觉得很多人不知道 Figma 其实也是一个基于浏览器的深度技术押注。他们对 Figma 的第一次路演还没有设计工具。他们的第一次演示是一个 3D 球掉进水里,在浏览器里运行的效果。
他们的路演核心基本上是,“浏览器有了一个叫 WebGL 的新能力”——WebAssembly 的前身——“有了这些东西,你第一次可以真正创建一个图形渲染引擎,然后在上面构建设计工具。但你必须从零开始写那个渲染引擎,因为没有什么现成的东西可以直接编译成 WebGL 之类的。而且如果你需要所需的性能等等,这要花我们好几年的时间,但如果我们做到了,我们认为这将彻底改变设计领域。“
押注 WebAssembly:WebContainer 的灵感来源
显然,我们现在都知道那个故事的结局了。回到 2016、2017 年,我和联合创始人 Albert 看到了同样的故事开始上演,只不过是在 Web 开发和开发环境的领域。具体来说,当时浏览器里落地了一些新东西,比如 WebAssembly、共享内存、Service Worker 等各种 API。我们当时就想,“哇,从理论上讲,应该可以用 WebAssembly 写一个操作系统,能运行 Node.js、NPM 以及你做 Web 开发所需的所有工具链。”
这将会意义重大,因为搭建开发环境对新手来说很痛苦。很多人在这个环节就流失了。你学写代码时做的第一件事甚至不是学写代码,而是学会怎么配置你的电脑才能开始写代码。如果你加入 Netflix 或其他任何一家大型科技公司,头一两个月就是在做入职培训,把那些东西跑在你的电脑上、搭建你的开发环境。我们当时就想,“如果这一切变成——你点一个链接,它就在浏览器里启动了——那将是一件大事。”
而且你看其他那些真正在 Web 上成功的生产力应用,它们都采用了这种计算模型,对吧?Figma,当你打开一个 Figma 文档时,并不是有某个云虚拟机为你启动来渲染文档。你拖动元素的时候,它用的是你的 CPU 和内存来执行工作。Google Docs 也是一样。这是唯一一种曾经扩展到十亿用户规模的模型。所以当你看云端 IDE 的时候——比如 Cloud 9 是最早的一个,大概 2009 年左右——这些产品一直以来运作的方式是:你的浏览器基本上什么都没做,当你打开它们的时候。每一个连接进来的用户,都需要启动一台云虚拟机,然后你的浏览器只是把你的按键操作发送到服务器,再把结果发回来。其他所有那些 AI 代码、文本生成应用类的工具也都是这样运作的,都在用云虚拟机。
端侧计算与云虚拟机的可扩展性困境
问题是,在小规模下这还行得通,但当你扩大规模的时候——地球上甚至都没有一亿台虚拟机可供租用。但有十亿台设备可以运行这些东西。因为我们在 Bolt 上看到的就是这种情况——如果你想做一个能扩展到那个规模的产品,你必须审视所有因素,然后说,“我们必须确保技术在体验上是最优的——零延迟、没有基础设施成本。“可以有一个宽松的免费层,因为服务器的另一个问题是,如果你提供免费层,人们会在上面挖比特币,会用你的服务器对别人发动 DDoS 攻击。所以你不可避免地要阉割这些功能、往回收。但如果一切都在终端设备上完成,这些都不是问题。
所以总之,WebContainer 是关键组件,而我们的困境在于,花了大概四五年的时间才把 WebContainer 建出来。之后那些年我们挣扎的是如何围绕它构建产品,因为开发者很喜欢它,但他们使用它的方式不会产生付费意愿。尽管我内心的极客多么希望这就够了——觉得做出很酷的技术就够了——但事实是,“不够。我们要做的是一家达到风险投资规模的公司。“所以这就是为什么我们在那段旅程的尾声处境艰难,一直在不断尝试。到了某个时刻,终于打中了,对吧?
深度技术押注的经验教训
Lenny Rachitsky: 这段历程中有很多非常有趣的经验教训,我觉得都是反直觉的。第一,你基本上是先做了技术,然后再去找要解决的问题。这通常是人们告诉你不要做的事。但在这种情况下,它成功了。
另一个有趣的收获是,感觉这类似于 AJAX 出来的时候,然后所有人都在说,“哇,你可以在这里构建全新的东西了。“所以这里似乎有一个教训就是,“如果有一项新技术使得某件大事成为可能,就在那个方向上深耕一段时间,看看会不会冒出什么来。”
还有一个教训就是,作为创始人,只要尽可能长久地活下去就好。因为你可能会找到行得通的东西。
Eric Simons: 说得都非常好,非常好。因为你说得完全正确。幸运的是,我和联合创始人在这之前已经做过很多不成功的创业。我们花了大量的时间——或者说尝试了二十次——在不同想法上反复试错。所以当我们有了确信的时候,我就觉得,“这似乎是一项会变得重要的技术。“Web 是世界上最无处不在的平台……我们脑子里的那个设想就是,“Web 是世界上最无处不在的平台,但它竟然没有——你不能用 Web 来构建 Web。”
其他每个平台,Mac 有 Xcode,Windows 有 Visual Studio,而 Web 什么都没有。我们当时就想,“至少 Google 应该把这个东西从我们手里买走。它似乎应该成为 Chrome 的一部分。“我们觉得,“嘿,这可能是一个巨大的赋能工具。“让构建全栈应用变得像使用 Canva 一样简单,这个愿景看起来真的很有吸引力。
创业生存法则:控制消耗率
但当你做那种高风险的深度技术押注时,你需要……这一点我们做得很好。我和 Albert 之前做的那家公司,一路自力更生做到了被收购,所以我们理解那种勒紧裤腰带过日子、自己出资创业的感觉。所以我们从中学会了如何保持低消耗率,如何大量尝试,如何让每一块钱的效用超出任何人认为合理的、可能的范围。这也是我们经营 StackBlitz 的方式。公司成立的前两三年我们没有融资,完全是自力更生。当我们真的融资之后,也几乎不怎么花钱。很大程度上是因为我们觉得,“我们需要做大量聪明的押注,而大量花钱没有意义。”
我想说一个普遍原则:在你看到需求拉动之前——就是人们争先恐后地从你手里抢产品之前——你不应该花钱。你的默认态度应该是”不”。当你去买软件的时候,你应该说,“我们是个小创业公司,能半价卖给我们吗?“你买的每一样东西都要谈判,把消耗率压到尽可能低,因为你需要尽可能多的尝试机会。因为你完全不知道结果会怎样。我认为总的来说,对于创业公司,在我看来这是正确的做法。除非你看到了即时的需求和拉动。
不过是的,我想补充的额外背景是,我认为我们在极度保守方面做得很好。那段时间——整个 2020 到 2021 年——正是狂热情绪和扩张团队规模被视为公司 KPI 的时期,是大家都在大力推崇的事情。很多人带着很大的情绪压力来告诉我们,“嘿,你们也应该这么做。“我很庆幸我们没有听那些建议,因为如果我们当时把公司规模扩大三倍、大幅提高消耗率,就不会有 Bolt 了。我们早就倒闭了。
创业者的非共识判断
Eric Simons: 所以我认为做创业者最难的一点就是……在某些时期,你必须做出不符合共识的判断。也许几年之后它会成为共识,但你必须对自己如何打好这手牌有信念。
Lenny Rachitsky: 这里有太多精彩的教训了,我觉得就是”活下来”这个理念。Dalton 上过我的播客,他是 YC 的合伙人,他有一句话叫”别死就行”。你们做的正是这件事——七年时间一直在尝试,直到某样东西奏效了。而且我知道你们推出 Bolt 的时候只发了一条推文,对吧?那就是发布时刻?
Eric Simons: 对,是的。
Lenny Rachitsky: 能不能聊聊发布之后那一刻,哪些信号让你觉得”好了,这次行了。这次不一样了”?
Bolt 发布后的爆发
Eric Simons: 第一天,推文的反响非常好。我们说,“哇,这是我们见过最大的发布日反响之一。“我记得第一天,我们大概增加了六万美元的 ARR。简直是疯狂。再说一下,我们当时是六十万的基数,所以一天就涨了百分之十。我记得我们的运维工程师,是他来提醒我的。他说,“各位,今天进了六万。这太疯狂了。“我说,“是是是,但这是发布日嘛。”
这有 TechCrunch 带来的热度,初次曝光的峰值,经典的创业公司——
Lenny Rachitsky: [听不清]——
Eric Simons: ——成名时刻,对。我说,“各位,听我说。“我在努力给团队降温。“这很好,但还有很多活要干。“然后第二天我们又增加了八万,差不多那个数,然后就一直在涨。而与此同时,我们发布的产品——我们用九十天做出了这个东西。我们用九十天做出了 Bolt。所以产品里缺很多东西。很基本的东西,非常基本的东西。我们确实在正确的 corners 上做了取舍,让它先上线了,但用户源源不断地涌入,说,“怎么没有移动端自适应视图?聊天消息怎么不能……”顺便说一句,我们在没有移动端自适应视图的情况下做到了两千万 ARR。说出来你们可能不信。就像 iPhone 一直到第五代才有复制粘贴功能一样。我们就是这个情况——没有移动端。你在手机上看,简直惨不忍睹。
类似这样的问题很多,所以我们不得不不停地修。而且我们是小团队,完全没有准备好应对不断增长的流量。每天发生的问题清单简直疯了。首先,我们在 stackblitz.com 上从来没有超过九美元的套餐。我们只有一个价格,九块钱。所以当 Bolt 上线时,我们说,“再说一次,我们不确定人们会不会喜欢这个,但九块钱买不到多少推理量。“于是人们四十八小时就烧完了九块钱。然后他们说,“我想多买点。怎么买更多?你们为什么不收我的钱?”
所以,不到一周我们就推出了全新的定价方案,可以升级,这套模式后来基本上成了行业标准。这个领域里其他所有玩家都抄了我们。在 Bolt 上线之前,Copilot 以及之前所有的 AI 产品,大家都想要 Netflix 模式——一个价格,无限量使用。问题在于,如果你这么做,你会希望推理成本相对低一些,因为你预期人们会大量使用。所以你没法做那种智能体式的体验,太贵了。
我们最终误打误撞发现的是:“好吧,实际上,人们愿意付更多钱。人们想为更多推理量付费,因为我们已经跨越了那个门槛——你能获得非常实实在在的投资回报。“你知道这给你带来了巨大的价值。总之,这是第一件事。服务器简直在融化。Anthropic 那边的 GPU 被我们用光了。Dario 给我发了封邮件,他说,“听我说,我们已经没有更多资源可以给你了。“那段时期,我们就在想,“我们该怎么应对……”简直疯了,持续了好几周。感觉就像《300》里那样,被一万人包围,而我们整个团队在拼命做所有的事。就十五到二十个人,什么都干。我的幕僚长和我一天百分之九十五的时间在做客服。总之就是这样。
二十人团队如何撑起爆发式增长
所以,是的,那是一段疯狂至极的时光。现在也还是,不过我们总算有一点点时间来适应这种规模了。通常来说,一家公司要增长到两千万 ARR,至少会有一年时间来逐步扩充团队。
Lenny Rachitsky: 通常要几十年。
Eric Simons: 是的。所以当时真的很艰难,我们会去问别人,“我们该怎么办?“拿回来的方案都是——花六个月,或者一年什么的。我们说,“这行不通。“但有趣的是,这恰恰就是全部的意义所在。至少对我来说,那种强度的挑战,是有难度的。但也是有趣的挑战,你懂吧?
Lenny Rachitsky: 哇,好吧。这简直是一刻不停的疯狂经历。你提到整个过程里你的团队大约是二十人,你们以这种疯狂的速率增长,只有二十人。这是怎么做到的?是什么让你们能以这么小的团队增长得这么快、这么多?你说的《300》的比喻很有意思,我猜拥有这些斯巴达勇士是其中的重要部分。到底是什么让你们做到了这一点?
Eric Simons: 是的,我觉得很大程度上……你看其他那些做代码生成应用的公司,很多都在服务器扩展上挣扎。而我们这边,既是”一夜成名”,也是七年积淀。所有这些东西,如果你倒回第二年,我们根本不可能——不可能达到这样的 DAU 增长、收入增长,什么都不可能。很大程度上是因为我们做出的技术,但更重要的是,是人。
人……在创业公司里,核心的五六个、七八个人能一起待五年以上,这很少见。在硅谷这算是相当罕见的。通常人们在一个创业公司待一两年,然后跳到下一个。你明白我的意思吧?而人员流动这么大的问题在于,你没法做那种需要长期投入的押注,就像我们做的那样。所以从一开始,这也追溯到我们之前自力更生做公司的经历——人更少,每个人脑袋里装的上下文更多。这就是我们的做事方式,而且我们对此非常坚定。
原因是,第一,你和任何人的沟通都能有很高的信任度,因为你知道他们掌握了大量上下文。不像有些人完全在暗处,待在公司的某个角落,什么都不……你懂我的意思吧?第二,每个人都有从头到尾把事情做完的自主权。没有什么政治博弈需要你去获得批准,没有……所以你看 Bolt 发生的一切,我们的工程师从头到尾——和遇到问题的用户在线上沟通,当场去修复问题,现场搭建 UI,直接上线。不需要牵扯团队里的其他人。
团队的信任与凝聚力
Eric Simons: 所以我觉得,这归根结底就是高信任度的结果,而且我们这些人过去合作得很愉快。也许这就是原因——也是唯一能让一个人在一家公司待那么久的原因。那些高压情境,我认为是决定成败的关键。对任何团队来说都是如此。所以我觉得,所发生的一切,实际上直接反映了创造和支撑这一切的人之间的力量与纽带。
Lenny Rachitsky: 对,我觉得这一点非常重要,你们这些人已经一起工作很多年了。大多数人不会有这个优势。你们在招人的时候,招聘初始团队的时候,有没有什么你们特别看重、但别人可能不太重视的东西?招聘新人的时候,你们优先考虑什么?是那种什么都能做的人吗?能做客户沟通、能做设计、也能做工程?
招聘哲学
Eric Simons: 对我们来说,甚至可以说我们对招我们的人也是一样的——我们在乎的是那些不在乎头衔的人,他们不在乎……不是说他们……当然,人都有职业发展轨迹之类的东西,但他们真正的动力就是做很酷的东西,把自我留在门外。他们在那儿是为了共同创造出伟大的东西,而不是只是跟着走,或者当那个”聪明但难搞的人”。我们招的大部分人都在欧洲。我们是一家完全远程的公司。我和联合创始人在湾区。有意思的是,2018 年的时候,我们还租了办公室什么的,每天通勤去上班。因为我们以为会招本地的人,结果过了一年我们想,“我们在干什么?我们两个人跑到一个办公室,十个人的团队,我们招的人全在欧洲,或者遍布美国各地。“到现在我们在湾区也就招了一两个人。
但我觉得我们基本上就是在找那种发自内心想做出好东西的人,对事物有好奇心。而且我们最早招的那批人,之所以能找到他们,是因为他们本身就是 StackBlitz 的用户。我们公司大部分招来的人,基本上都是从我们的社区里来的。所以当我们想招人的时候,就发条推文说,“嘿,我们在招工程师,“然后就会收到私信之类的。
大概就是这些我们看重的基本素质吧。
Bolt 的优先级决策
Lenny Rachitsky: 我还想再问几个关于 Bolt 的问题,然后我想把视角拉远一点,聊聊未来的方向。我们来聊聊优先级排序。我猜你们肯定被各种请求淹没了——就像你描述的那样,产品上线之后,各种请求铺天盖地。就像你说的,一百万月活用户。我简直无法想象你们收到的功能请求数量,再加上你们自己知道想做的那些东西。你们到底是怎么决定优先做什么、实际去构建什么的?
Eric Simons: 有很多东西,人们甚至不知道是可能做到的,所以大家不一定会明确地提出需求。于是就有这么几种情况,我们凭直觉判断——“嘿,至少没有大量的人在要求这个。但我们觉得这会是一件大事。”
最好的例子就是上周上线的原生移动应用支持。从反响来看,这是我们发布过的最大的功能。但在公司内部,甚至有些人都不太确定——“这个,我不知道……用户在喊的是其他那些东西啊。“确实,这始终是一种平衡:有多少精力用来处理各种待解决的问题,又有多少投入到新能力上。但当时就是觉得,“这个我觉得很重要,“于是我们往桌子中间推了一些筹码。而且赌对了。那个体验简直不可思议,现在每天都有成千上万的移动应用被创建出来,而以前根本没有。这会怎么改变局面呢?我是说,现在有些小企业,以前根本不会去做一个 iPhone 应用。完全没有意义,太贵了。现在不一样了。
所以就是这类事情——“嘿,我们应该在这上面押注。“但我觉得最好的类比可能是,就像在餐厅当厨师。一方面有来自顾客的反馈——“这个东西不好吃。“另一方面是,“嘿,我们一直在研发一些有趣的东西,这个味道……我不知道,我觉得你会喜欢的。我觉得这是一道杀手级的菜。“所以你得在这两者之间找到平衡。
而且我觉得,这在很大程度上就是多年经验积累的结果。如果倒回十年前,我基本上不会……不会有这么多年被自由市场反复教训所培养出来的直觉。你得自己摸索出一套判断力,我想这是最好的说法了。
Lenny Rachitsky: 再稍微展开讲讲,你们有一个固定的节奏来决定做什么、发布什么吗?比如每周开会之类的?因为我知道答案很可能就是——“基本上就是持续的混乱,不断地在灭火。“我知道大部分时候确实是这样,但你们有没有某种流程,来决定做什么、怎么同步、怎么和团队协作?
Eric Simons: 我们每天都开会。基本上整个团队都会上同一个电话,从头到尾过一遍——
Lenny Rachitsky: 视频会议?比如 Zoom?
Eric Simons: 对,每天太平洋时间早上八点,我们至少开一个小时的 Zoom——
Lenny Rachitsky: 每天?全公司?
Eric Simons: 基本上全公司。对。
Lenny Rachitsky: 哇。一个小时。好的。
Eric Simons: 对。我们就把所有事情过一遍。我觉得随着团队规模增长,我们可能会开始拆分成不同的同步会之类的。但是每天把所有人放在同一个房间里这件事——很多人会抱怨,说……在 Twitter 上你会看到有人说,“哦,这是对所有人时间最昂贵的浪费。“但我想说的是,“没错。但这里面沟通的保真度损失是 0%。每一件事,每一天,都从头到尾被审计,从头到尾被讨论。”
所以在这种极端增长期,你希望沟通中的信息损耗尽可能接近零。这就是我们的做法,尤其是 Bolt 上线之后。我想 Bolt 上线后的那一周我们就决定了:“每天,直到我们挺过这一阵,所有人每天都要上同一个电话,从头到尾过一遍。“再说一遍,这也是为什么更多人需要更多上下文、更少独立脑袋——公司里的每个人都了解公司其他所有事情的进展。所以人们可以独立做出决策,而这些决策默认来说,正确的概率比错误的更高。
Lenny Rachitsky: 这太有意思了。我之前从来没听过这种做法。尤其是对于像你们这样在成长中的公司来说。你们居然这么做,超级有意思。
Eric Simons: 我觉得我们不会一直这样做。
Lenny Rachitsky: 对,对。当然。
Eric Simons: 但,是的。
Lenny Rachitsky: 不,但我觉得这一点非常值得注意——它确实管用,而且在你们这里确实奏效了。那你们讨论完之后,东西放在哪里?路线图放在哪里?怎么做规划?公司工具栈里都用什么工具?
工具栈与产品流程
Eric Simons: 工程方面,我们大量使用 Linear。产品路线图方面,我们用 Notion,在 Notion 里写 PRD 类的东西。设计方面用 Figma。不对,实际上我们现在用 Bolt 做很多设计和原型了,你能想象。但我觉得工具方面没什么疯狂的,没有什么特别复杂的。我觉得我们会投入更多,尤其是当你开始把人们从每天同一个电话里拆分开来的时候,这些东西才真正变得重要。因为那时你就没有一个时间窗口可以动态捕捉那些本来不会被提出来的事情了。
Lenny Rachitsky: 我太喜欢你们还在用 PRD 了。我甚至喜欢你用了这个词。现在有很多人说,“哦,我们有了 Bolt,有了这些工具,不需要 PRD 了。我们直接做个原型就行了,就这样。“能不能聊聊你们为什么仍然觉得 PRD 有用,以及你们在 PRD 里写什么内容?
Eric Simons: 除非我们在做一个非常复杂的东西,否则我们倾向于让 PRD 保持很轻量。我喜欢只放最少量的上下文,确保每个人在同一频道上,确保我们正在做的那个功能的关键成果在到达终点时会呈现出来。因为当这些文档变得非常冗长的时候,你看一眼就觉得,“天哪,太多东西要消化了。“问题是,当它被交到开发、设计或者其他环节时,很多人会直接略过。然后就开始像滚雪球一样越积越多。所以最好尽量保持简单。至少这是我们的做法。而且很多时候,有些 PRD 的内容就是:“这是一个 Bolt 的链接。”
Lenny Rachitsky: “这是它可能的样子。”
Eric Simons: 对。而且不只是”看起来”的样子——“这是一个基本能工作的 demo,能让你感受到它实际上会是什么体验。“因为如果一张图值一千个字,那一个活生生的 demo 就值一百万个字。你能感受到它。它是真实的。我们现在看到很多采用 Bolt 的企业,就是在用这个做高保真原型,因为用 Bolt 做真正的原型现在更快了。以前,成本太高了。那种”让我们做个原型,工程师来写一个 proto……”的想法——那要花很长时间,也很贵。而现在用 Bolt 在代码里做这件事,得到一个真正能运行的软件产品,比在 Figma 里拖拽框架做一个静态版本还要快。
用 Bolt 构建真实产品
Lenny Rachitsky: 那我们来聊聊这个。企业用 Bolt 究竟做到了什么程度?原型是大家对这类工具的普遍想象。我知道目标不仅仅是做原型,而是要构建全规模的产品。我设想长期来看,是 Salesforce、Atlassian 那种大规模的公司。有没有一些用 Bolt 构建的产品案例,其完成度会让人们感到惊讶?
Eric Simons: 有的,尤其是……当你从零开始做的时候,你可以用 Bolt 来构建……比如说 Salesforce。最早注册 Bolt 的人里有一个叫 Paul 的,他是个创业者,不会写代码。三周内就构建了一个 CRM,内置了 AI,接了 Stripe 做支付等等。他之前从一家外包公司拿到报价,要三万美元,耗时六个月。他用三周就做完了,而且他在 Bolt 上只花了三百美元。所以你看,这是……他已经在靠这个赚钱了。这是他的创业起点。对吧?
Lenny Rachitsky: 好的,所以他构建了这个产品,而且在卖。有人在付费使用。
Eric Simons: 对。对。
Lenny Rachitsky: 哇。
Eric Simons: 这样的案例还有很多。如果你看从零开始的项目,百分之百可以。在当前前沿模型的能力下,你完全可以构建生产级软件。你不可能一步到位,但你会花几天、几周的时间。但成本降低的幅度——三万对三百美元,便宜了 99%。六个月对三周。这完全是数量级级别的交付速度提升。这些数据对我们接触到的那些在构建全栈应用的人来说很有说服力。人们去 Upwork 上找外包,拿到五千美元的报价,最后五十美元就搞定了。用这个东西能做到的事情简直疯狂。
所以在另一面,很多已有的公司也有非常合理的使用场景,其中从零开始的部分可以快速搭建。一个典型的例子是公开网站。营销页面、落地页等等。人们正在采用 Bolt 来驱动这些,而不是用 Webflow,比如说。因为这比 Webflow 更简单好用,而且能和公司现有的设计系统集成,营销人员不需要会写代码就能更新内容等等。但对于产品开发团队来说,对于已有的软件公司来说,最常见的用法是加速产品开发流程,但不是说整个从零开始、“嘿,我们在 Bolt 里构建整个产品”那种方式。
Lenny Rachitsky: Bolt 能和现有代码库集成吗?还是暂时还不行?
现有代码库的适用边界
Eric Simons: 对,我们实际上可以在 Bolt 中打开代码仓库。你可以直接用 Bolt 操作你的代码库。不过这取决于你的项目配置。确实也有公司在上面搭建营销网站,或者管理后台之类的。我认为这个使用场景未来会有越来越多人采用。不过这些 LLM 目前还不够强大,取决于你的应用有多大。如果你的项目有一千个甚至更多文件,目前还无法获得非常可靠的体验。但一年后我们再聊,我猜答案会不一样。所以这取决于应用的规模和体量——如果太大,你更多是用它来快速原型开发,纯粹加速产品开发流程。如果不算太大,那你就可以完全在 Bolt 里搞定。
Lenny Rachitsky: 这很有参考价值。你觉得目前 Bolt 的主要局限是什么,哪些方面大家需要知道”暂时还做不到,未来可能会有”?听起来,如果你有一个非常大的现有代码库,可能还不是最合适的工具。还有什么大家应该了解的?
Eric Simons: 我觉得这大概是主要的一条,因为如果你有大型代码库,你需要的是 Cursor 这样的工具,而且你得是一个真正的开发者才能去编辑那些东西。除此之外,就像使用 Photoshop、Figma 或者单反相机任何生产力工具一样,你需要一定的学习投入,学会如何使用这个工具,才能真正解锁它最大的能力。
我们在 Bolt 上看到最成功的非开发者用户是那些非常优秀的 PM。这些人通常对技术运作有足够了解,而他们的工作本身就是指导开发者如何改进产品,把需求规格写得清晰可执行,减少沟通中的信息损耗。当你想”怎样才能最好地和 AI 开发代理协作”时,本质上就是这样——你需要擅长定义范围,帮助它调试各种问题。做一个顶尖 PM 的能力模型,和用好这些文本生成应用、代码生成工具的能力模型,有着巨大的重叠。
PM 角色的价值重估
Lenny Rachitsky: 我很高兴你说了这一点。这正是我一直想表达的,我也为此写过一篇 newsletter。因为当这些工具刚出来的时候,很多人说”PM 这个角色完了,不需要了,我们可以这么快速简单地构建东西,PM 还有什么用?“但我完全认同你的看法。现在难的部分变了——构建变得容易了,难的是”我们到底该构建什么?我们能不能清晰表达出想要的东西?“然后,“我们有没有足够的判断力知道这是否正确、是否好、是否真能解决问题?“再往后就是增长的问题,这也是 PM 一直在思考的。所以我完全同意。基本上,我觉得 PM 是在这个新世界里最有优势蓬勃发展的角色,我的播客很多 PM 在听,他们会很高兴听到这些。
Eric Simons: 毫无疑问。这也是 Bolt 增长过程中我们……因为我们之前是面向开发者的产品,所以预期用户百分之百都是开发者。但我们不断看到越来越多的非开发者在使用它,到现在我们 67% 的用户都不是开发者。当我开始和这些人交流的时候,一开始我觉得很奇怪,心想”这是怎么回事?“但后来突然就想通了:“哦,这将会改变一切。整个软件行业的格局将被彻底重写。”
因为今天公司组织构建软件的方式,将会完全改变。PM 是真正理解到像素级别什么才能构成出色产品体验的人。而我作为一个开发者自己也知道,他们不得不反复催促开发者,才能把东西调整到应有的样子,细到最小的层面。而如果你往前看一年、两年、五年,情况会变成:PM 将会自己”写代码”——加引号的写代码——而不是写一个 JIRA 工单然后等开发者来做。开发者将专注于那些 LLM 擅长不了的、真正有智力挑战的任务,同时 LLM 也会辅助他们完成这些工作。而 PM 则可以直接自己动手做修改。
组织架构的变革
让我感到震撼的是,这件事还没有被定价——没有任何一家公司把这一点算进去。全球软件公司的组织架构也没有反映出这种变化。这一切都将彻底改变。至少那些赢家公司的组织架构会彻底改变,他们构建和交付产品的方式也会彻底改变。这一切已经开始了,而现在只是起点。
Lenny Rachitsky: 我想顺着这个话题深入聊,但在此之前,我想补充一点,如果我说的不对请你纠正。我觉得当我们谈论 PM 的时候,这也适用于创始人,特别是有产品思维的创始人——
Eric Simons: 百分之百同意。
Lenny Rachitsky: 非常类似。另外我觉得有一点也很重要,如果你同时具备工程能力和设计能力,你会更有优势,这只会加分。但如果你看产品、工程、设计这个三角关系,感觉 PM 的技能将是未来最重要、最有价值的。当然如果你还能写代码、还能做设计,那就更好了。
Eric Simons: 完全同意。而且对我来说,这是最令人兴奋的组合。我觉得 PM、设计师和非技术背景的创业者,这是最让我兴奋的事情,因为一个全新的市场正在被打开。有史以来第一次,这些人可以直接写代码、自己构建产品,把他们的愿景直接转化为软件。这将改变一切,而且已经在改变一切了。
Lenny Rachitsky: 你说组织架构会发生变化,你具体怎么设想的?就是工程师会变少吗,还是未来的组织架构会是什么样子?
Eric Simons: 好问题。我猜未来某一天会有 Gartner 的分析报告,不管过多少年,会说”最优秀的公司是这样组织的”,会用某个术语来定义。但我觉得我们会看到开发者可能被从大量用户界面相关的工作中抽离出来,至少总体趋势是这样,除了那些最复杂的部分。你会看到设计师和 PM 真正地冲在前面,负责打磨那些体验。可能还会配备一个开发者来审查代码,确保方向正确,审查那些 PR 等等。而且我觉得,工程师们——就像你说的,具备工程能力不会是劣势,只会让你效率高得多。
Eric Simons: 但我确实认为,前端工程师能获得的杠杆将会变得极其惊人,而且只会越来越强。所以我预见到,前端工程师的数量会减少,而产品和设计人员会增多,可能配备一两个工程师之类的。你会看到更多类似这样的小团队组合。我觉得这大概就是未来的趋势。
Lenny Rachitsky: 这让我想到,我们之前请过一位来自 OpenAI 的研究员上播客。她其实最开始是在 Anthropic 做前端工程师的。她说一旦看到 AI 对前端工程能做什么,她就想:“我得换个方向了。” 于是她转到了研究岗位,因为她预见到那个角色可能会消失。这和你说的完全一样。
Eric Simons: 是的。
给下一代的建议
Lenny Rachitsky: 那我问你这个问题——我不知道你是否已经有明确的观点了——假设你刚有了孩子,假设你的孩子将来要上学了,假设你的孩子马上要上大学了。你对他们应该学什么技能、进入什么领域,以及应该避开哪些现在很热门但将来可能不再热门的方向,有什么想法吗?
Eric Simons: 关键是要学会如何利用这些 AI 工具。我觉得不一定非要……可能对编程的基本原理有个了解是很好的。
Lenny Rachitsky: 就是技术基础,理解系统怎么运作、编程怎么运作。
Eric Simons: 没错。但这不一定要……因为回想一下,如果 Bolt 当年就存在的话——Albert 和我经常对彼此说这件事——从一开始做 StackBlitz,我们就在打造我们自己 13 岁时希望拥有的东西。而且说实话,后来我们做的一切也是一样。尤其是 Bolt。我是说,如果那时候有这个工具,我不确定自己还会那么深入地去学编程、去做工程师。我们当初入行的原因是我们有想要做的产品想法、商业想法。而编程只是为了实现这些想法的必要手段。
当然,我认为人们应该追随自己的内在兴趣。如果有人真的对计算机底层原理感兴趣,对编程语言、Spark、编译器之类的东西着迷,那就去做吧。我觉得那些东西仍然会是相关的。我不确定我们是否会真的达到……到时候看吧,但如果到了 AGI 的程度,就是说我们再也不需要思考任何问题了——
Lenny Rachitsky: 对,每次聊到这个答案都一样。那我们还卖什么?
Eric Simons: 是啊,如果真到了那个地步,那就有点……我不知道,我不确定。但我觉得,至少从我能预见的未来五年来看,人们仍然可以找到自己的专业化方向,真正深入下去。但我觉得你应该带着一个想法去学,而不是”我要去学计算机科学,因为毕业肯定能找到工作”。
我只是觉得这普遍不是一个好的……这么说吧,我和我的联合创始人都没有上大学。我的联合创始人上了一学期就退学了,而我压根没去,因为当时我们就在写代码,做外包项目,在赚钱。而且那意味着要背上一大笔——你知道,读完四年州内学费就要负债十万美金,在伊利诺伊大学。当时好像十二万美金吧。
果不其然,这确实是一个很大的问题。人们被……2010 年代初或 2000 年代末,社会上有一个普遍的观念,认为上了大学就能找到一份高薪工作。但事实证明对很多人来说并非如此。而且我认为这种情况还会持续。不过,也不是要劝阻别人去上大学,而是你必须非常确定:“我真的想要这个,我想成为这个领域里最优秀的人。“你明白我的意思吧?
Lenny Rachitsky: 我觉得这段话翻译到你孩子身上就是:“甚至可以考虑不上大学。”
Eric Simons: 除非他们自己想上。我觉得对 18 岁的人来说,这个要求太高了。甚至不是 18 岁,是 17 岁,因为你那时候就要开始申请大学了。这对一个收入为零甚至为负的年轻人来说,是一个巨大的六位数债务承诺。除非你真的有很强的信念,否则在线上免费探索和学习的成本是零。
AI 时代最重要的技能
Lenny Rachitsky: 我想回到你认為最重要的技能这个话题上,让我试着把你说的几点复述一下,这些我非常认同。
所以我觉得,如果你想在 AI 越来越能替你构建东西的世界里取得成功,我听到的是:要善于发现人们需要什么、想要什么,需要解决什么问题。要善于向 AI 工具清晰准确地表达你的需求。现在有人说”你不需要成为出色的提示词工程师,不需要在提示词上下功夫”,但我觉得这越来越重要了,因为你告诉它一件事它就去构建了,如果你表述清楚,就能省下大量时间。
所以核心是:善于发现人们需要解决的问题,善于把问题表达清楚并要求一个明确的解决方案,然后学会怎么推动增长——这个需求会一直存在,因为 Bolt 不会替你去……我可以看到运营付费广告之类的事仍然需要人来做。这会是一个持续的需求。
然后我觉得还有一个”解卡”的环节,就是帮助 AI 摆脱困境,我觉得这可能就是工程技能越来越发挥作用的地方。你对这个技能怎么看?
Eric Simons: 完全同意。我们大概两周前发布了一个叫 Bolt Builders 的项目。它基本上就是苹果商店里的天才吧(Genius Bar),当非开发人员在 Bolt 上构建东西时,他们会遇到一些角落和缝隙,AI 就是搞不定。我认为这种情况在可预见的未来还会持续存在。这就是我们的判断,也是为什么我们启动了这个项目。
Lenny Rachitsky: 这些是真人来帮你。
Eric Simons: 是真人。是我们认证的人。所以在接下来的几周里,Bolt 会有一个按钮,你只需点击”帮我连接一位认证专家”,就可以和他们实时聊天,帮你解卡,每小时收费五十美金左右。然后你解卡了,继续写提示词。我还是觉得这一切都很棒。工程师可以专注于有难度的挑战,而不是那种千篇一律的”再做一套 CRUD 应用”。他们可以做调试——调试是很有挑战性、也很有趣的。去做那些智力上有刺激性的任务,而那些复制粘贴、反复重复的东西,错误处理那些——让 AI 去干那些活儿吧。
AI 优先,人其次
Lenny Rachitsky: 这可能是一个全新的职业,至少目前来说,就是帮 AI 解卡。我觉得随着时间的推移,AI 会越来越强,也许以后就不再需要这些人了。但我很喜欢现在这种模式:AI 在先,人在后。而不是人来构建东西,AI 辅助。当 Copilot 刚推出的时候,它就是”好的,给你这个函数的小建议”,而现在完全反过来了。“给你全部代码。“然后,“哦,这里我不知道怎么办了,帮帮我们。“于是就变成了一个人的建议。这不是很有意思吗?就像,人类的 Copilot,整个关系颠倒了。
Eric Simons: 完全同意。是的。真正令人震惊的是,我认为 Sonnet 确实是第一个扭转了这个等式的模型,因为真正推动这一切的是我们,还有早期的 Cursor,以及所有那些其他产品。爆发式增长恰好从 Sonnet 上线的那一刻开始。实际上我们差不多正好在一年前就尝试过构建 Bolt,用的是当时的前沿模型。花了一两周时间来开发,但就是不行。输出的代码不够可靠。它总是会出问题,要么应用是坏的,要么看起来很丑,诸如此类。后来在五月我们提前看到了 Sonnet 的东西,我们当时就觉得,“哦,好的,我们应该把这个项目从架子上拿下来正式启动,因为可能就是它了。”
果不其然,事情的发展完全就是这样。但这就是关键所在,在引擎盖下面,正在发生的事情是——LLM 编写生产级代码和构建真正美观、真正运行良好的应用的能力,已经跨过了一个非常关键的门槛。它还不完美,但确实出现了这种从零到一的质变,“好的,所以现在,是的,AI 成了第一主体”,然后你偶尔才把一个开发者插进来,而不是反过来。
Lenny Rachitsky: 我之前真不知道。我没有意识到这一切如此大程度上是建立在 Anthropic 的工作之上的,尤其是 Sonnet。你是说那是第一个能够把代码写得足够好的模型。
Eric Simons: 是的,毫无疑问。
Lenny Rachitsky: 哇。
Eric Simons: 毫无疑问。绝对如此。
Lenny Rachitsky: 太有意思了。仅仅那一个模型所释放出来的收入、商业和电商的量级,简直疯狂。我之前完全没意识到。
Sonnet 为什么改变了游戏规则
Eric Simons: 确实如此。回想起来,我之前提到过,我们在 StackBlitz 从未做过 AI 产品,虽然这个念头很诱人。当 ChatGPT 上线的时候,所有人都开始往自己的产品里加 AI。但我们就是看不到一个能真正带来增值的明确切入点。所以我对那些——你知道,很多人说”AGI 在 2023 年就会到来”,诸如此类的说法——并不是特别看好。我的意思是,当时各种言论满天飞,而我的态度是,“我只是不确定它是否真会像人们说的那样快。“在某种程度上,这个判断是对的。
但我当时没有深入思考的是,如果 AI、如果 LLM 要在某个特定垂直领域变得更强,那会是哪些领域。以法律为例,你想做出最好的判例法 LLM。问题在于,法律不是确定性的。法官的裁决取决于当时社会对事物的看法、政治因素、陪审团。有大量的变量使得它不是确定性的。所以你很难创建出足够可靠的训练数据,来产出非常好的结果。你也不能凭空编造案例说,“理论上法官会这样判。“因为,这根本行不通。
软件是确定性的。当你写完代码按下运行,它要么跑得通,要么跑不通。这就是 Anthropic 真正洞察到的关键。他们就是深耕了这个方向。他们做的事情就是,基本上对你能构建的每一种类型的应用进行排列组合,做强化学习,然后启动成千上万的核心去跑。就是大量地构建训练数据,做强化学习,让他们的 LLM 成为世界上最擅长构建美观、可靠应用的模型。我极其看好。从技术上讲得通为什么,在所有领域中,LLM 在写代码这件事上的进步速度会比大多数其他应用领域都快得多。原因很简单——因为这是一个可以极度确定性的东西,而且每秒可以被排列组合成千上万次。
所以我认为更宏观的趋势是——Sonnet 已经把所有人都惊醒了。Google、OpenAI,所有——所有人现在都在猛攻编码领域,因为重新改写软件世界秩序的市场机会有多大?是数万亿美元的量级,对吧?整个世界都运行在软件之上。所以我认为,从最宏观的视角来看,这也是我们为 Bolt 融资的原因。这个机会看起来极其清晰,你可以把人们说的那些炒作和什么什么的 hype 分离开来。当你从技术上拆解开来,这一切是说得通的。这里正在发生的事情是说得通的。对我们来说,我们非常庆幸自己处于一个有利位置,能够帮助人们乘上这波 LLM 创新的浪潮,而且这波浪潮还会持续到来,从而让更多人能够构建出更加疯狂、更加了不起的软件。至少这就是我们的世界观,关于宏观层面正在发生什么。
Lenny Rachitsky: Sonnet 是什么时候发布的?已经有一段时间了吧?
Eric Simons: 我觉得他们正式上线应该是六月。
Lenny Rachitsky: 所以从六月到现在,这是 AI 编码最差的状态了,而它已经这么好了。而且自从去年六月以来,他们还没有发布过新模型。这告诉我们,一旦他们发布下一个模型,事情会以多快的速度加速。而且正如你所说,所有人现在都在猛攻这个领域,因为他们意识到”我们在编码这块落后了”。所以,哇,这将会变得非常疯狂。
Eric Simons: 我同意。是的,再次说,没有什么博客文章把这些全给我们 laid out。就是这种——
Lenny Rachitsky: 基本上就是你们发现代码变得非常好。
Eric Simons: 从那以后,我们一直在把所有这些碎片拼凑起来。所以有一种侦探破案的刺激感,“这里到底发生了什么?”
Lenny Rachitsky: 真的吗。
Eric Simons: 是的。你懂我意思吧?
Lenny Rachitsky: 就是拼凑你们观察到的 Anthropic 发布的东西,是这个意思吗?从侦探的角度来说,你们注意到了什么?
Eric Simons: 一方面是这个,另一方面是 Bolt 的影响,我们有非技术背景的人在使用它。他们是怎么用的?为什么这样做?然后,我们在这期播客中谈到的所有内容,都是九个月 R&D 的成果——看到结果,然后”什么?“然后深挖,再做一件事,然后又是”什么?“它就是不断发生,因为这条路没有任何先例可循。
Lenny Rachitsky: 这就像一种人类学——如果这个词用得对的话——就是在观察。听起来像是一种涌现式的发现,而不是你们先有了策略,“好的,我们要做这个,“针对某个用户画像来发布这个产品。
Eric Simons: 是的,完全同意。是的。我觉得这是最好的表述方式了。就是这样。所以这一切非常有趣——
Lenny Rachitsky: 去观察,并且成为其中的一部分,我想象是这样。
Eric Simons: 是的,是的。
模型的进步才刚刚开始
Lenny Rachitsky: 而且我觉得当人们说”好吧,这些不过是玩具,只是原型而已,没法和你的现有代码配合,也没法扩展”的时候,有一点很重要,就是我们刚才讨论的内容。这还是在去年六月的模型上就能实现的效果,而所有人都在研发下一代更前沿的模型,会让这一切变得更好,而且很快就会到来。好的。太棒了。
Bolt 的下一步计划
Lenny Rachitsky: 我还有几个问题来结束这次对话。一个是,Bolt 接下来会有什么新动向?在节目播出之前会有哪些酷炫的新功能上线?也许是刚播出之后,也许是三月中旬?
Eric Simons: 好的。所以到时候——我回头要跟我们的工程师说,“我在播客上说了这个——”
Lenny Rachitsky: “我已经承诺了。抱歉啊各位。”
Eric Simons: 这其实……我发现自己在播客上总是嘴巴不严,我工程师就说,“你怎么能告诉他们……”我就说,“那你们就得更快交付啊,现在。得把它变成现实,“对吧?
Lenny Rachitsky: 或者我们让 AI 来搞定。
Eric Simons: 不不。是的。但我觉得对我们来说,还是一样,我们看到很多产品经理、设计师、创业者等等在使用 Bolt。所以我们真的在考虑更好地与人们目前用来做这些事情的工具进行整合。Bolt 不会取代它们之类的。如果你在一家公司工作,你怎么把这些东西和你现有的业务、现有的产品和代码库整合起来?因为这是我们经常被问到的问题,就是”我怎么打开我的——“比如前几天我们在跟一家风扇公司聊,他们说”我们怎么把我们的生产代码库放进去,那个代码库有二十年了?“我说,“你不要。这些工具都不行。这不是你应该做的。在你的场景下,这是用来做快速产品开发的。“
从 Figma 到全栈应用
所以我们即将推出的功能——我对这个特别兴奋,我们已经做了一段时间了,而且我们和一家叫 Anima 的公司合作来共同完成。基本上就是,在任何 Figma URL 上,当你看着你做的设计时,只要在那个 URL 前面加上 bolt.new 然后回车,它就会把那个设计吸进 Bolt 里,然后直接变成一个全栈应用或移动应用,开箱即用。
Lenny Rachitsky: 这太天才了。
Eric Simons: 是的,它——
Lenny Rachitsky: 太棒了。
Eric Simons: 会很疯狂。是的,我是说,真的很好玩。因为不管你是开发者还是设计师,或者什么的,把它拿过来然后变成一个真正的编码应用。关键是,一旦进了 Bolt,你就可以继续在那里面做 prompt。你说,“好,在这里再加一个页面。“所以那些你需要像素级精确设计的地方,你可以做到,而且它是一比一转换的。它还会拆分出资源素材。Anima 从 2017 年就在做这个,从 Figma 到代码。他们拥有世界上最好的 agent 来做这件事,他们是排名第一的 Figma 插件。所以在 Bolt 里,它就是能直接用。是深度整合的。
Lenny Rachitsky: 所以就是 bolt.new/ 加上 Figma 的设计 URL。
Eric Simons: 对。你只需要做这个。
Lenny Rachitsky: 太棒了。
Eric Simons: 而且在 Bolt 里,我们还会在聊天框里放一个小的 Figma 图标。所以如果你直接去 Bolt 里,你可以点击它然后粘贴 URL 之类的。但就是这样。从 Figma 到全栈应用,一键完成。真的是这样。太疯狂了。所以这个功能挺酷的。
Slack 里的 AI 开发者
然后我们还在做的另一个功能是和 Slack 的整合,因为通常当你在公司里讨论 Figma 链接之类的东西时,你都在 Slack 里。所以你在进行这样的对话,“嘿,我们真的应该给这个加一个页面来做什么什么。“所以我们实际上在做一个 Bolt Slack 机器人,它的角色基本上就是充当你团队中的一个开发者。所以你可以在一个帖子里说,“嘿,我觉得我们应该做一个首页。""好的。@Bolt,你能快速搞一个出来吗?“然后它就会去拉取你 Slack 里到目前为止的整个对话历史,不管是帖子里的还是什么的。
你会看到,“好的。所以你大概想让我拿这个 Figma URL,我可以自动转换它,“这得益于我刚才提到的那个功能。“我可以直接把它转换出来,然后你想让我给它加一个页面,再做这个事情。明白了。我这就去做。哦,这是 URL,你可以打开它然后继续做 prompt。“你明白我的意思吗?就像是有一个开发者或者什么人来帮你把这个东西搞定。你只需要说,“去把这个东西快速做出来。”
所以这两样东西,我觉得还是一样。你开始想象公司将会如何改变他们目前做产品开发的方式。甚至就是,“嘿,我们需要搭一个营销网站。就是这个。“我们就说,“Bolt,你能做吗?""没问题。我这就去做。“你明白我的意思吗?这就是为什么我对这两个功能特别兴奋,因为我觉得它们会很受欢迎,大家会很兴奋的。希望如此。但愿好运。
Lenny Rachitsky: 这些功能太棒了。我喜欢 Slack 那个部分,因为当我想到 agent 的时候,大家总是在谈论 agent。对我来说最简单的理解方式就是一个 Slack 机器人,就是 AI 能像一个人一样在 Slack 里和你对话。我很喜欢你做的正是这件事。这是一个巨大的功能——就是,“嘿,你现在有了一个工程师,可以帮你去构建东西。“
工程团队的工具栈
让我沿着这个方向问一个问题,我一直想问但之前忘了。就是,你的工程团队用什么来构建 Bolt?我猜应该是大量使用 Cursor。目前 Bolt 自己在多大程度上参与了构建 Bolt?还有没有其他他们觉得有用的、值得提及的工具?
Eric Simons: 好问题。是的,我们确实用 Cursor。我们的团队大量使用 Cursor。我们也大量使用 Bolt 来做产品开发流程,非常多,我们在用它。我们基本上就是我描述的那个流程,如果需要像素级精确的东西,我们会去 Figma 做。然后通常我们会把它拉进 Bolt 里,因为我们现在已经能用那个整合功能了。所以把它拉进 Bolt 然后说,“嘿,去加上这些东西”之类的。或者直接说,“嘿,这是我们 UI 的截图,去做什么什么。”
开发者还在用的其他 AI 工具。我觉得这些是主要的。我是说,我想我们有 Claude 和 ChatGPT 之类的订阅。但我觉得对于开发来说,Cursor 是最主要的。
Lenny Rachitsky: 是的,很有意思,工具其实很少。AI 工具有那么多,但有趣的是人们最终真正在用的工具却那么少。就是 Cursor、Claude、ChatGPT,然后可能再加一个工具。比如 Bolt。
Eric Simons: 是的。完全同意。
给新用户的建议
Lenny Rachitsky: 好的。差不多最后一个问题。假设有人第一次打开 Bolt。你想象一下,你可以坐在每个第一次尝试 Bolt 的新用户旁边,在他们耳边悄悄说一个成功使用 Bolt 的技巧。那个技巧会是什么?
Eric Simons: 这个嘛——因为我们有很多不同类型的用户。我想你说的是产品经理或设计师之类的——
Lenny Rachitsky: 就说产品经理吧。这个好。这里很多听众都是产品经理。
给产品经理的建议
Eric Simons: 我会说,像你写 Linear 工单或 JIRA 工单那样跟它对话。这就是我的建议。就像跟团队里的开发者说话一样去跟它沟通。意思是,在关键的地方要具体明确。而在那些你可以放手让它发挥的地方,就随它去。你可以直接说,“嘿,让它好看点。“它做得不错,实际上你只给它一个感觉,它就能做得非常好。所以总之,我觉得对产品经理来说就是——你具备这个技能。你知道怎么做。就把它当成你的同事,你的开发者同事。
Lenny Rachitsky: 我喜欢这个建议。因为这些工具太容易上手了,你进去,说一句话,然后,好了,你有了一个网站。代码写好了,搞定。而我从你这里听到的是,花点时间去打磨你的需求。你可能很想直接上手,“好,给我做一个 CRM”,然后你就被卡在那个第一版上了。然后你说,“哦,好吧,我不是这个意思。“所以听起来你的建议是,花些时间精心构思你的需求,把你想要的想清楚。
Eric Simons: 没错,完全同意。尤其是当你对自己要构建的东西有清晰的愿景,而且是比较复杂的东西时。我给大家一个建议,如果你第一次用 Bolt,不知道该让它做什么,也没有想法——让它给你建一个个人网站。这里有某种魔力。你把 LinkedIn 上的内容复制过来,粘贴你的 LinkedIn 简介和工作经历,选中文字、复制、粘贴。“我需要一个网站,我叫某某某,这是我的 LinkedIn 经历。我最喜欢的颜色是蓝色,我喜欢狗。“然后点一下粘贴,对吧?
Lenny Rachitsky: “让它好看点。”
Eric Simons: 对,差不多。然后你就可以点击部署了。如果你还没有自己的 .com 域名,现在就可以有了。你拥有了一个真正个性化的网站。我觉得那种时刻会让人产生一种感觉——“哦,好吧,哇。“这基本上是零次调整,99.999% 的情况都是一次搞定。你得到了一个你之前没有的、漂亮的个人网站,放在 Wix 上做的话至少要花你一个小时,甚至更久。这让你体验到——“哦,好的。所以如果我真正花时间去把这件事想清楚,写一份 PRD,然后逐步把它输入到这个东西里,那 possibilities are endless。“
AOL 办公室的故事
Lenny Rachitsky: Eric,这次对话在很多层面上都太疯狂了。我有太多需要消化的东西,我想很多听众也是。也许作为真正最后一个问题——我看过一个故事,说你在创办 StackBlitz 的时候,甚至更早之前,你在 AOL 的办公室里蹭住。因为你有一张还能用的门禁卡。能讲讲这个故事吗?
Eric Simons: 对,那是我以前最出名的事。那发生在 2012 年,我当时 19 岁,所以已经是很久以前的事了。我现在大概 33 岁了吧。
Lenny Rachitsky: 追诉期应该已经过了。
Eric Simons: 对对对,没错。但确实有一段时间我在想,“我得做出比住在 AOL 更了不起的事情。不能让这个成为我的标签。“但现在人们会说,“哦,等等,你就是那个 StackBlitz 的人,你也是那个 AOL 的人?”
所以我第一次来硅谷的时候,我们在做一个 K12 教育创业项目,那时候 YC 只给你两万美金,所以 YC 有一个分支叫 Imagine K12,是 Jeff Ralston 搞的——他后来好像做过几届 YC 的 CEO。后来 Imagine K12 在 Rob 参加之后就并入了 YC。不管怎样,他们在 AOL 办公楼里有办公空间。当时 AOL 正试图重振公司,他们说,“嘿,我们应该让年轻的创业血液进来,那就把办公空间租给年轻人吧。“基本上就是这个意思。
所以我在那里,后来我们的钱花光了。两万块在硅谷撑不了多久,大概三四个月的时候我们就慌了,“天哪,怎么办?“我当时每周去 AOL 办公室好几次,因为我们有门禁卡可以进去找投资人的办公室。然后我发现——这里有沙发,有吃的,有可以用微波炉加热的拉面,还有带淋浴的健身房,甚至可以洗衣服。所以我就想,“我不知道,也许在我想办法的这段时间,我就先住这儿吧。“于是我就在那儿住了下来,大概住了四五个月。我住在 Palo Alto 的 Page Mill 和 El Camino 交界处那个 AOL 总部里。
然后我确实蒙混了一阵子,因为那些保安——他们是 12 小时轮班制。所以晚上值班的那些人……我基本上每天都在写代码,从早到晚。晚班的保安就觉得,“哇,这家伙真拼命。“早班的保安也说,“这家伙工作太努力了。“我跟其中几个还成了朋友。后来我觉得可能是因为还有很多斯坦福的学生,他们好像在某个过道里都摆上床铺了,事情开始失控了,所以他们开始严查。然后有一天凌晨四点左右,一个保安进来了,把我赶了出去。
我是芝加哥来的,在这儿谁都不认识。那时候我想,“我在湾区一个人都不认识。“我去了附近一家 Starbucks,但还没开门。我就睡在外面桌子边上。然后我联系了项目里的另一个创业者,我说,“你有沙发吗?我觉得我现在确实需要一张。”
后来媒体知道了这件事,变成了一个全球性的新闻。但最疯狂的是,我每天只花一美元。我的日均开销是一美元,那时候。
Lenny Rachitsky: 那一美元你花在什么上了?
Eric Simons: 那个时候麦当劳还有一美元菜单。真的。所以偶尔去吃个芝士汉堡什么的。对,那就是极致的艰苦创业。
Lenny Rachitsky: 你当时严格来说是无家可归的。从无家可归,到历史上增长最快的创业公司之一。Eric,多么不可思议的一段旅程。这是你人生中如此有趣的一个节点,也是科技行业如此有趣的一个时刻。无论将来发生什么,我相信你都会非常成功,但这确实是这段旅程中一个如此有趣的节点。感谢你抽出时间跟我们分享这一切。
Eric Simons: 有这种视角总是好的——你创办公司应该保持一种心态,就是你做这件事是为了好玩。所以不管怎样,看到这一切会走向哪里,我都很兴奋。一定会很有意思。
尾声
Lenny Rachitsky: Eric,最后的问题。大家如果在网上想找到你、想就你分享的内容跟进交流,去哪里找?听众们怎么才能帮到你?
Eric Simons: 当然可以。网站是 bolt.new,Twitter 上我是 @ericsimons40,我们的 Twitter 账号是 @boltdotnew,注意没有句号,就是 B-O-L-T-D-O-T-N-E-W。我很想听听大家的想法。再说一次,我们从前来尝试这个工具并给出反馈的用户那里学到了很多东西。上线后的第一次会议中,我们就已经不再是这个工具的最佳使用者了,从那以后一直是如此。所以我非常喜欢听大家说接下来想看到什么功能、这个工具怎么帮助了他们,以及在哪里遇到了问题,哪些地方需要我们去修复。我的邮箱是 Eric@stackblitz.com,Eric 拼写用 C。无论通过 Twitter 私信还是邮件,都欢迎联系我。
Lenny Rachitsky: 太棒了。Eric,非常感谢你来参加节目。
Eric Simons: 太好了,非常感谢邀请我,太开心了。
Lenny Rachitsky: 大家再见。
感谢大家的收听。如果你觉得这期节目有价值,可以在 Apple Podcasts、Spotify 或你喜欢的播客应用上订阅。也请考虑给我们评分或留言,这真的能帮助更多听众发现这个播客。你可以在 lennyspodcast.com 找到所有往期节目或了解更多关于这个节目的信息。
下期再见。
术语表
| 原文 | 中文 |
|---|---|
| AJAX | AJAX(Web 异步通信技术) |
| Albert | Albert(Eric Simons 的联合创始人) |
| ARR (Annual Recurring Revenue) | ARR(年度经常性收入) |
| Bolt | Bolt(产品名称,保留原文) |
| Canva | Canva(产品名称) |
| Cloud 9 | Cloud 9(早期云端 IDE 产品) |
| Copilot | Copilot(产品名称) |
| CRM | CRM(客户关系管理系统) |
| CSV | CSV(逗号分隔值文件格式) |
| Dalton | Dalton(YC 合伙人) |
| Dario | Dario(Anthropic CEO Dario Amodei) |
| DAU | DAU(日活跃用户数) |
| DSLR | DSLR(数码单反相机) |
| Dylan Field | Dylan Field(人名,保留原文) |
| Eric Simons | Eric Simons(人名,保留原文) |
| ERP | ERP(企业资源计划系统) |
| Evan Wallace | Evan Wallace(人名,保留原文) |
| Expo | Expo(公司名称,保留原文) |
| Figma | Figma(产品名称,保留原文) |
| Gartner | Gartner(IT 研究与咨询公司) |
| Google Docs | Google Docs(产品名称) |
| GPU | GPU(图形处理器) |
| JIRA | JIRA(项目管理工具) |
| Lenny Rachitsky | Lenny Rachitsky(人名,保留原文) |
| Linear | Linear(项目管理工具,保留原文) |
| Netflix | Netflix(公司名称) |
| Netlify | Netlify(服务名称,保留原文) |
| Node.js | Node.js(JavaScript 运行时环境) |
| Notion | Notion(协作工具,保留原文) |
| NPM | NPM(JavaScript 包管理器) |
| OneSchema | OneSchema(产品名称,保留原文) |
| Photoshop | Photoshop(图像编辑软件) |
| PRD | PRD(产品需求文档,Product Requirements Document) |
| React Native | React Native(技术名称,保留原文) |
| Service Worker | Service Worker(浏览器 API) |
| SFTP | SFTP(安全文件传输协议) |
| Squarespace | Squarespace(产品名称,保留原文) |
| StackBlitz | StackBlitz(公司名称,保留原文) |
| Stripe | Stripe(在线支付服务) |
| TechCrunch | TechCrunch(科技媒体) |
| Upwork | Upwork(自由职业者平台) |
| Visual Studio | Visual Studio(产品名称) |
| WebAssembly | WebAssembly(技术名称,保留原文) |
| WebContainer | WebContainer(技术名称,保留原文) |
| Webflow | Webflow(网页设计工具) |
| WebGL | WebGL(浏览器图形渲染技术) |
| Wix | Wix(产品名称,保留原文) |
| Xcode | Xcode(产品名称) |
| YC | YC(Y Combinator,创业加速器) |
此文档由 AI 分片翻译(translate_long_document)