From Nearly Shutting Down to $40M in 5 Months

We were thinking about shutting the company down and returning the remaining capital to investors. At this point, we were four years in, we were under 1 million in annual recurring revenue

from nearly shut down to 40 million a RR. In just five months,

we decided that there'd be one more big experiment and had this experiment gone poorly, that very likely would've been the end to the company.

We launched it with just a tweet, it immediately went viral. I have never experienced and probably will never experience another product launch like this.

Over 70% of the people signing up didn't have any previous experience coding. Just gonna make a calculator app. So there we go. We've got a calculator.

I remember when I was,taking Python classes and doing, this is one whole semester's project

at this point. It's. Extremely clear to me that anybody can do it. If you don't believe me, just DM me on Twitter and we will work it out together.

Hello everyone. Welcome back to Two Set ai. Today I am talking with Alex Berger, COO at Bolt New, This is the AI platform that went from nearly shut down to 40 million a RR in just five months. Alex was employee number seven.

Correct.

And you are the first non-engineer hired.

Am I right?

Yeah. and thankfully they welcome the engineers welcomed me in with open arms.

Alex, thanks for joining.

Thanks, Sam.

It was definitely a team effort, but, appreciate the kind intro.

Hey everybody, I'm Alexander Berger. and yeah, I would describe myself as a generalist and I've known, the CTO and CEO of Bolt, our co-founders for a long time. we actually met,for me was the early days of Silicon Valley.

back in like 2012, they ran a hacker house in Palo Alto and I came by one of their legendary parties that they used to throw, got acquainted with the founders, and the rest is history. yeah, it's been a awesome journey to get to join the team. and as you mentioned, yeah, I was coming in as the first non-developer into a very technical.

Product focused team. and actually back when I joined in 2020, we were a developer tool. So I do always like to say I was the first non-developer to join because, being the first non dev at a dev tool company, you're stepping in with some heavy hitters.

yes. I wonder how you navigated that environment at time.

Yeah, I think part of it is just, a great team. and something that's so cool about our company is that I was employee number seven and all seven of those people, including myself, are still at the company today. and. It is really, credit to the technical folks there that they were just super inviting and, I've always been, someone that can bridge the gap between the business needs and the technical.

So I think that helped. And, it was honestly a pretty natural, natural fit. Once I got to know everyone on the squad. Hmm.

I know how it feels for technical founders. go-to market is really a mystery.

A hundred percent. And to be honest, it was largely a mystery to me as well at the time, how I had actually ended up joining the company was I reached out to our CEO, Eric and I was talking about some of my past experience and what I was looking to do with my career, really just looking for some advice.

And one of the things that I shared with them is I'd done a lot of different roles at companies before from it, finance, operations, customer support, but I hadn't been involved in any of the teams. On the go to market side that are really responsible for generating revenue. And so sales and marketing was, top of my list in terms of wanting to add that to my skillset.

And I was really just looking for some advice on where I might go to develop those skills. But next thing you know, I joined the company as with the remit of being our first enterprise. Sales rep. and at the time we were selling on-prem developer tools into very security conscious enterprise companies.

And yeah, I got to work with our CEO and we figured it out. built that motion from scratch. I learned a lot. and, uh, you know, there's still parts of it that are, I think, a mystery to everyone, but it was a really awesome experience to be able to make the, first sale ever of a product and,bring that win back to the team that built the product.

How much was that for sale?

13,500, which was,for anyone who sold on-prem software knows that's not the dollar amount that you want to be selling. It's a lot of work to onboard a new customer and set them up, where you're installing your app in their, their architecture.

And realistically, we were trying to target $54,000 as our minimum because just of all the work and fall. But, I was eager to have a quick win for the team. So we did that deal. but it was followed up, not too long after with us the first, six figure deal. that really got things rolling.

yes. Congratulations. It must be hard. I've sold to Enterprise and I know on the outset, right? We all downplay ourselves, right?

Yeah. and when you're selling to a large organization, it's not just one person usually that's involved in making the decision.

So you've really, and when you're a new company, you don't have existing relationships with these companies. You really have to dig in and understand how different companies make decisions and who you need to pull in. yeah, definitely, challenging, but, very interesting learning experience, both on how to sell things, but also just how big companies make decisions and purchase things, which is really helpful for anybody that's trying to build a business to business tool.

Absolutely. I would love to hear more about your tactics of, doing that, later.

before we start, I would love to ask this question, to all our founders building Gustav, can you explain what both does Like how would you explain it to an elementary schooler or maybe college student,

This helps us, your audience understand your product better.

Absolutely. So I think also, I can do a little show and tell here so we can see exactly what this looks like. the starting with the elementary school, description, I would say that bull is a tool that you can tell it what you want to make, whether that's a website or an app or a game, and it'll make it for you pretty straightforward.

And that largely does capture what the tool is and. What does that look like? it's basically a box in which you can type in your idea and the tool will build it for you.

to give a quick demonstration here, I thought of something that might appeal to a elementary kid and typed out this prompt.

just gonna make a calculator app that looks the same as on MacBook, except it has a few interesting, fun features. one is we'll have a little, race car track around the edge of the screen, and the larger the numbers that you're working with, the faster the car will go. And then also just have some, animated confetti effects to make things a little bit more fun.

The other thing I'm gonna do, this is optional, but if you really want visual accuracy in the design, a good thing to do is attach a screenshot. So I just took a screenshot of my calculator And pasted that in, so that can just help the tool. Nail the design a little bit better. So I'll, send that off to, to build for a moment. and while we let that cook,

I can say, okay, so how would I explain this to someone in college? I'd take it up a notch, probably, a little more advanced. So if you think about what does Bolt, how does Bolt fit into the ecosystem of tools you might be familiar with?

If you were gonna go make a website a year ago, you'd probably think of a tool like Wix or Squarespace and where it's this template based tool for making websites. And drag and drop things around and resize things and set colors and such. obviously with Bolt you just described your idea, but importantly it's also unbounded in terms of what you can create.

So you can of course create a simple, like small business website like you might on Wix, but you can also integrate real deep functionality into what you're building. you can integrate, you can have like login, right? You can collect payments, you can have users upload files and store those.

You can create accounts and store databases and information that you're collecting from your users. So really you can build. Like a software service app, anything you could build like an Instagram clone, a Twitter clone, or just a website that has a lead capture form. And you could build that from scratch if you wanted.

Really excitingly, you can also integrate via APIs with almost any other tool in the world. So you not only can you build your own software, you can make it connect and interact with other people's software. that's probably like the college level explanation.

But, as we can see here, after we type our prompt in the tools planning out,the different steps and going through them, and you can see we've reached the final step here where it's testing and building the project, the plan's completed. And now we're going to get a preview, in the right hand, pane here of the application that we just built so we can check it out.

So there we go. We've got a calculator, it's got the iOS, color scheme, and we can see we have a race car going around the edge. So now let's see what happens if we, let's see. Multiply 500 by 500. There we go. the car speeds up and we get a little confetti effect. Now we can maybe try some division.

We get the confetti effect again. There you go. that's just a very simple kind of nonsense example. You, I don't know why you would necessarily wanna create an app like this, but it could be anything, right? this is effectively a game at this point. It's not, it's not that incredible of a game, but,you can make the car go faster with math,

yeah, crazy.

I remember when I was,taking Python classes and doing this is a, this is one whole semester's project

and that happened while I explained what bull is to an elementary school kid. It's absolutely wild how much this speeds up your ability, to take. Any idea you have in your head and get it onto the screen.

and that's really the exciting opportunity, that exists, for folks that are, like professionals in this industry. there's of creating digital products. these tools are really starting to redefine how, how these digital product teams operate. And, so I think as we kinda.

continue on here. We'll probably get a little bit more into the details of that. But at a high level, I would just say if you're on a digital product team, this tool can be used to massively accelerate the way you're collaborating with your teammates and getting your products, scoped out, designed and built, at work.

It does look very powerful. the now do, would you say this is the starting point or this will carry you through your journey of product development?

So that's the whole question. and. We are seeing some very interesting things developing in the industry. And my thinking on this has even shifted a bit since you and I first connected and planning out to do this, interview.

So these tools didn't exist 12 months ago. it just, or 14 months ago. that's when we launched. We were the first ones to bring a AI powered coding tool into the browser. And the reason why they didn't exist earlier is because it wasn't possible. the AI models were not good enough to reliably write code that would.

Not have bugs and actually exhale. So last summer, Enro came out with sonnet 3.5 and that was really like the go point for this type of tool where the models were good enough and they've gotten a lot better since then. so it's really exciting as we see each evolution. we are absolutely at a point in history where depending on what you're building for many use cases, this tool can get you all the way across the line to a application that is deployed and, oh my goodness.

And used by your users and customers.

Yeah. this is amazing. it's hard to believe that this is happening,

It's super cool. for me. I've had this perspective for a long time, since really my first job after college.

It dawned on me that in some sense, all jobs are really the same thing. You just have an objective, a set of tools and capabilities to go after that objective and then limitations or constraints on achieving that objective that are, imposed on you by reality. And that's just only becoming more and more true, especially in this space of digital products.

Where a tool like Bolt, if you have the idea, you are much less likely to be limited in realizing your idea by some detail. oh, you're not a programmer or you're not a designer, so how could you make a digital product like product that those constraints are going away very rapidly, which is awesome to see.

I think I should send, both to my mom, she recently trying to build an architecture related application and she was chatting with Chad GPT and I should just send this to her. you should try this. this is making your life so much easier.

It'll make it for you. Yeah,

a hundred percent. and we've seen all kinds of folks build interesting applications with this. I know our CEO's mom made her first website using the tool, and my wife made a website for her small business. So plenty of people that you know, are not technical minded by nature, can absolutely be successful.

And a great way to get started is if you've been discussing an idea with Chat GPT or another one of these AI chat bots, just ask it to summarize and write out a prompt that would be appropriate to put into one of these tools. And you

off to the races at that point?

yes. and, and you mentioned that you just got promoted to the Chief operating Officer at Boat.

Congratulations.

Thank you. Thank you.

what does your day-to-day actually look like now?

Yeah, so it's, it shifted a lot, as we talked about a little bit, I came in and worked in sales and spun up that department and then moved through most of the other departments at the organization from people ops and marketing and customer success, and a little bit in product.

And, doing my best to fill the gaps at the time and then hire, experts, area experts behind me to take over and build those functions. got to a point this year where we had. Leaders taking on a lot of those core functions. And so there wasn't an immediate need to go zero to one, some new department that didn't exist.

And so that's when I started working towards this new role.

And the way I've been describing is my job is to ensure that the CEO's vision for the company turns into tactical execution across all of the teams. My weeks are looking like spending time with our functional leaders.

oftentimes it's working sessions, right? we're a startup, we're not, Google or something. So there's of course, strategic planning that does need to go on.

But a lot of what's really required to push a company at our stage four is getting things done. So working sessions are with the leaders, is a big part of the job, the other day I was sitting down with our interim head of marketing Enrique and we're working through some collateral that we're making for the sales team, and just thinking through and ideating.

And I also am running finance, There's, reviewing the month end p and l and looking at how that ties into our product strategy and keeping track of margins and customer retention and communicating that out to the other leaders and figuring out how that fits into the strategy. also, we're working on a lot of partnerships right now,that's another area that I'm owning and we're working on some, exciting partnerships with very large, companies that have major distribution into enterprise.

So having those conversations, laying the groundwork for co-marketing and co-selling activities there.

So it's, the stereotypical, it's always different, but it's a lot of, essentially. diving into some of the tougher problems that we're working on, cracking with the leaders and trying to get us to that, to the next screen of the game, essentially.

it doesn't sound to me any easier from zero to one. It just sounds like more work.

Yeah. Yeah.

I was thinking, You're transitioning out of some of the functions that you actually build from scratch and then maybe less work. No. not less work. It is actually more work.

Yeah. certainly don't come work at a startup if you don't like solving a continuous chain of very tricky problems. 'cause that is effectively what the last five years has been. it's like just constantly unwinding the next puzzle. And, the puzzles keep changing, and the shapes of the pieces and the way they lock together and the size of the pieces and the quantity of the pieces.

It's,it's definitely not boring, that's for sure.

I remember seeing that you guys had a hard time, like at least one or two years ago. What happened?

Yeah, so, you know, when I,joined, we were a developer tool company and we were called Stack Blitz, and the tool was basically the same thing as Bolt.

It was a code editing environment in the browser, but there was no ai, so you needed to write the code yourself and There was a few issues with that product. The most obvious one being most people do not know how to write codes. So the number of people that you could sell that product to was limited to software developers.

there was also some more, that's a little tongue in cheek, but we didn't have, a hundred percent product market fit for. Businesses massively adopted in the open source community and in a lot of open source library documentation sites and everything.

But we were trying to sell to these design system teams in the enterprise, which are these really important teams that are actually very relevant today to what we're building with Bolt as well. Where they make like the, what I think of as the Lego, the visual Lego building blocks of an enterprise application.

Because when you go into a Google product, all the buttons look the same, all the search boxes look the same. So there's a team, or in the case of Google, multiple teams that just build out those component libraries for reuse. And so we were selling to those teams and they loved us, but there wasn't really a reason to roll out our tool to everyone at a company,

this is niche.

Yeah. It was extremely niche. And we, so as a result, as we were coming up into the final stretch of 2024. We were thinking about shutting the company down and returning the remaining capital to investors because we had pushed that product as far as we could, and we were not getting the commercial success that we needed to justify continuing on.

And so at that point, we decided that there'd be one more big experiment that we were gonna run.

And, honestly when Eric brought the idea to me initially for Bolt, I was not that convinced. this is really, before it was these, now everyone knows AI coding is a major part of the ai,revolution that's happening.

But at the time, I was not sure that this was gonna be the golden bullet, right? That was gonna save the company. But at this point, we were four years in, we were under 1 million in annual recurring revenue. And we had a team of about 15 people at that point in time. And so we decided to run this last experiment with Bolton and hook AI into our product and put it out there.

we launched on October 3rd, 2024 and had this experiment gone poorly, that very, very likely would've been the end of the company. yeah. Thankfully that's not how things turned out.

that's crazy. it was just a year ago.

Yeah, it's been a quite a whirlwind year for sure, for the whole team.

what did you do differently? Do you think the product worked at the time? It's the right timing. And what about the go-to market function you were building in the meantime?

Yeah, so timing is so important and I tell everybody with, our story, we were lucky enough. Enough to stay alive long enough until the core technology that we built became commercially relevant there.

fundamentally what allows us to, to run our product, and this is deep in the weeds, nerdy stuff, but we developed this technology called Web Container that allows us to run the whole development environment in your browser tab. So that's, always been a core aspect of our product instead of spinning up.

a remote server, abased

server,

everything. Yeah,

yeah.

All that. we just run the code locally inside your machine, which is very helpful for security performance, scaling. we can support millions of users without having to run millions of servers

Sounds

like

a hack.

It's a huge hack. the way I sometimes describe it is like when we're talking to AWS or gcp, I'm like, you guys are renting out your servers to your customers, but what we're doing is actually renting out their own computers to them. So it's a genius business model.

Right?

But that technology allowed us to do stack blitz, right? And we had this niche tool, but it wasn't really until we could sell. A development environment in the browser to everyone that the technology reached. Its real potential in terms of commercial viability.

So honestly, before I would say anything about go to market, the number one thing I would say is if you're running a startup and you are not 100% sure that you have product market fit, you figured out how to deliver value to your customers that they cannot live without and that they will pay you significant amounts of money for.

Be very careful with your cash because we ran the company very lean. Mm-hmm. And we're very just focused on traditional good company building, preserving cash, having sustainable margins, not blowing money on flashy things that don't really, help us improve the product and deliver value to our customers.

So it was really that philosophy that even allowed us to survive four years without passing another million dollars in revenue. And had we not survived till that moment, no. Nothing we would've done on the go to market front would've mattered. 'cause we would've had to shut down. Yeah,

so true.

Four years of resilience. how did you guys keep the mindset going? You joined on the first year or second year Mark?

Let's see. I would've been in 2020. They, it was about two years in to the company already. so yeah, it's a good question. Like one of the things is just the early team.

people that are very product focused. nobody joined the team. thinking of it like a get rich quick scheme, which I think is rarely the correct mindset, when you're trying to build something durable. Maybe never. You could argue, I'm not sure.

no, I'm not gonna argue with you on that one.

Yeah. So that was, everybody came in knowing we're here to build a product that delivers value and that's gonna be challenging. and everybody has love for that game of building a company. and I think, I gotta say a shout out to Eric and Albert, our co-founders, for keeping the original squad together all the way.

Through this day. when you look around and

Yeah.

People you've been with since the beginning, still there with you in the trenches. for me, that's been a huge,

yeah.

a huge thing that kept me going through the tough times because you really do, I don't believe in this idea that a company is a family.

Like we're not a family. We're a company here to conduct business, but we are a team. And knowing that your teammates have been with you through all the ups and downs, of the company, that is another thing that has contributed a lot to just our resilience and surviving up to the point where we finally had the right product

had deck through.

Yes. Yeah.

And did it happen fast? And what happened? What was the moment?

Yeah, that's the thing is we've done a lot of product launches together and. Some of them, you think it's gonna be the best thing ever. And you put it out there and it's a total flop.

And other ones you put it out there. And this was more our experience with what success looked like in the past was you'd put it out there, you'd get a big spike in new users, as you talked about it and pushed it out on social media and everything. then you'd have that big spike and then it would be a good fit for certain people and they would stick with the product, but the excitement would die down and you would have a tapering off of new users joining until sort of the next, launch.

I wasn't sure which one of those to expect when we launched Bolt, we. We launched it with just a tweet, a tweet thread on our Twitter. That was it. We were thinking of it as a bit of a soft launch. Put it out there, get some feedback from our core stack Blitz fans, and then go from there in terms of trying to push the word of mouth further outside of the circle.

but that is not at all how it played out. I have never experienced and probably will never experience another product launch like this. We put it on Twitter, it immediately went viral. YouTubers started making videos about the product, and people were talking about it, telling their friends, and it just exploded immediately.

Like we could not have ever imagined. so yeah, that it was a very different experience than any product launch we've done before.

who sent out a tweet? Was it a company or It was Eric?

that's a good question. I can't remember whose account we put it out on. usually we'd pick one or the other and then amplify from the other account.

Yeah.

at the time did either Eric or the company had a lot of following on Twitter. I was just wondering like what the virality

Yeah. I think at the time probably between those two accounts we're talking roughly 40,000 followers. and most of them frontend software development.

People involved in design systems or open source software development. And so we really thought that we were just launching it to some developers to give us early feedback. But very quickly what we saw is over 70% of the people signing up didn't have any previous experience coding. So it immediately broke out of our, our bubble of sort of software development fans and started getting shared into more I guess broader tech Twitter.

That's very interesting journey. and you think tho those 70% new users are actually coming from that tweet?

At the time, yes. Now we're obviously 14 months in, and so we've expanded our efforts to get the word out there. But yeah, in the early weeks there was not anything else that we were doing.

And it was just people were retweeting and next thing we know, we had a bunch of new fans of the product from a whole different kind of backgrounds than our previous customers.

did you, do any, any sort of analysis on that viral tweet? maybe you want to recreate that kind of virality again.

Yeah. and we try to kinda follow the similar approach we did for major product launches. it's a formula that we'd done before. So the approach we take is just very straightforward educational tweet thread, right where it, the first tweet is by far the most important.

And you'll just see that if you look at any viral tweet thread, the first tweet gets 90% of the views. And then no matter how good your thread is, the rest of it, the views are gonna drop off significantly. So it's very important that you get the most succinct and compelling version of your idea in that first tweet.

And Eric and I spent probably 12 hours at least bare minimum, constructing this tweet thread. So we will agonize over, your character limited on, on, on Twitter before it gives you the dreaded show more to read the rest of the tweet. Yes. Which we do not like. So we will agonize over those first 240 characters and do many iterations trying to get just the most clear and succinct version of the simple idea that we want people to walk away.

from the thread with. And so we'd done this, and and then I'll continue down in, in the rest of the thread, it's about going deeper for the people that we're curious enough to click. Then what, what are the components that you want to educate them on? And ideally you have a nice gif or static graphic for each of the points that you're trying to make.

And this is a format that we've done in the past and that worked really well and we've continued to do for new product releases, later on. And it's nice because you can then turn that into a blog post. You can turn it into an email. Yeah, you can expand. You basically have the most succinct version of the narrative 'cause you're so limited on space.

And then you can expand that out for other channels.

you mentioned 12 hours. Is that per tweet or every product launch tweet? 'cause it's so important that you spend most time on.

Yeah, I would say. That's the first tweet gets by far most of the time. So if we spent 12, we honestly probably spent more than 12 hours on that tweet thread, but let's just say it was 12 hours.

Yeah. We probably spent, nine hours on the first tweet and getting that messaging perfect and just making sure we had the right bullets for the follow on tweets. And then usually we'll just pop out the follow on tweets pretty, much quicker. 'cause at that point, if someone's already opened your tweet thread, they're engaged, they're interested.

Yeah. They now are there just for the information and to learn. I think it's the first tweet and the second tweet I guess is very important as well. and then, after, if you weren't able to get the every aspect of your. Your idea into 240 characters. I'll allow you a second, two 40 to hit a second point.

And then from there, it's basically the only people reading beyond that are already hooked, already interested. And you might as well just educate them on what they need to know to succeed with the product after that point.

so today, are you still using Twitter to be your major launch platform?

We are. I do think that'll be shifting a bit in the new year as we're really honing in our focus on business to business. but the reality is that Twitter and ai, Twitter sees it first. And so the, it's still where we announced our new features. first I think, the most, one of the most recent features that we dropped was, for example, supporting the new model that Anthropic came out with Opus 4.5 and.

Even though, we are focusing on B2B, people who work at companies are also on Twitter and a lot of learning about what's new in AI happens on Twitter. So that's, that has remained so far, our primary like feature launch channel.

your, does your first tweet have any, like videos you mentioned, gif or maybe graphics, but do you have videos as well?

we haven't been putting videos on the first tweet. we put 'em in the thread and no hyperlinks in the first tweet for sure. I don't, the thing is the Twitter algorithm changes all the time, but as of now, you get algorithmically punished for putting a URL into the first tweet. So maybe Nikita will.

Rescue us from, uh, duffing hell of URLs in the first tweet. But for now, you definitely, want to save those for lowered doubt in the tweet thread.

Wow, these are great tips. And when you're launching on Twitter, are you doing relaunches,the same thing, you're just announced it multiple times?

We haven't done that. but that could be an area that we've overlooked because frankly, like some of our favorite features of mine that we have in the product, a lot of people don't know about them. So like one, one feature we have, we call a visual inspector, where you can, a little cursor, you can actually select a particular element of the I love that.

Yeah, it's really helpful if you're making fine grain tweaks to a particular component in an app or on a page. I think only 5% of our users use that feature. Maybe we should, maybe that

needed, maybe.

Yeah. Yeah.

I remember I used actually earlier version of bolt, to bold applications.

I've used, deep, site. I don't know if you know about deep site it's a hugging face, so it's a hugging face space by deep seek. And then you can also build like landing pages. And part of it for fine tuning is I just using the user chat, but sometimes it will modify other parts that I don't want it to change.

the back and forth is the part that. Very tiring for you.

Yeah. and that's where, some of these lesser known features really can come into play.

yes.

Visual inspector. the other one is like planning mode, right? Because the worst feeling is when you send off a prompt and then you realize that the LLM has misinterpreted your intent and goes off and yes, it, doesn't implement the plan exactly how you would've liked.

Maybe it changes some things you didn't want it to change. And so that's another feature that we should perhaps relaunch on Twitter is plan mode. That it'll, you give it your prompt, it has to generate a plan which you can then review and revise before approving it. which gives you a little bit more of a tighter fist of control over your LLM

that's mimicking the real workflow from the PM team.

I feel, and it makes sense,

Hundred percent. Yeah. And I think that's a pattern actually that comes through in a lot of AI tools. where they're trained on human generated data, that came out of human generated processes. And even the whole practice of how we build digital products, the way the teams are constructed, and the different handoffs between product design and engineering.

it's a human process. And so when you're working with an AI tool, it's been trained on the outputs of that human process. Oftentimes, the more you can mimic that or borrow words and phrases from the discipline, the better outcome that you're going to have.

Absolutely. Yeah. how do you feel about the term vibe coding?

Is that what both is?

I don't like the term. And, but it is the term, right? That has caught on that absolutely describes our space. I liked it at first. I did think it was cool at first, but now I've gotten pretty, pretty, less hot on it. And I think that it's gotten a bad rap for the idea that you just throw half baked prompts over the wall and let the AI figure it out and you don't really have responsibility to be implementing your plan and using your, control over the tool to get the outcome that you need.

and The term I really prefer is software composing or software composer for the person doing it, because I really think that better captures what's actually happening, you are more like orchestrating and composing the app from various inputs. And it's your job to ensure that the tool you're using has the correct context loaded in at any given time to accomplish the task.

And I think software composing captures that, better than the term vibe code than

vibe coding. is this new?

it's not new. it, it was one of the original terms, I think, Andre Carpathy came up with Vibe coding,

right?

And. Riley Brown is where I think software composing came from.

very early on,I remember seeing software composing, and I like that from the beginning, but it hasn't caught on yet.

Yeah, it's like writing music, isn't it? in your mind, who's qualified, who can write, compose, software

at this point? It's extremely clear to me that anybody can do it. And if you don't believe me, just DM me on Twitter and we'll work it out together.

you'll argue with me on this one,

and the argument's gonna be easy for me to win because.

All you have to do to start is just describe your idea and paste it into the box, right? I think as the concept gets more and more complicated, the skills required to create software, uh, come more and more into play, right? So with the little demo we did at the beginning, an elementary school kid can go make a game.

Now they can make a game like you think of those old mini clip style games, 100%. kids in elementary school are making those all day every day. If you are trying to make something significantly more complicated than. You're going to need to do more than describe it in the way that elementary school kid would describe their game.

that's correct. There is a skill set that can be developed and needs to be developed to compose software at a professional level, but the skill is not software engineering.

What is this skill? I'm curious.

so I'd call it software composing and what is that made up of? It's multifactorial and I, I think people are still figuring this out, but if I could highlight just a few points of what I would focus on.

one you want to get extremely good at describing. Very specifically what it is you're trying to build in the language of digital products, right? So if you don't know the language of digital products, the beauty is AI will teach you everything that you need to know. hop into your favorite AI chat and just talk to it.

Ask it to educate you on the words and phrases that product managers and designers use specifically tho. Those are the most important words and phrases for you to understand as you're getting started with composing software. So you'll come across things like. talking about user interfaces and there's like a hero section, right?

Talking about the part at the top and different ways of talking about menus and navigation. Like a pancake menu is that little thing on mobile that looks like three stacked pancakes. So if you want one of those, you say pancake menu

and now you know the term for it, right?

Yeah. and so it really is like in fantasy novels where it's like in with magic, I feel like it's always you gotta know the name of the thing and that gives you the power over it.

Yeah. That is actually how it works in this space. And there's also a lot of, for the specifics of the visual interface, there's these concepts like padding and margin and alignment and font weight, and all of these design terms, right? Gradients.

those are used to be entry barriers, you know, it's intimidating because every domain is speaking a different language.

Totally. and that's something that you, that is a real skillset that you, if you want to be serious about building software as a non-developer, whether you're a founder, entrepreneur, product manager, designer, executive at a company, and you want to be able to put your ideas on the screen, there's definitely things to learn,

and those words and phrases are a great place to start. yeah. I was just gonna say, if you're, if you want to take it further, right? That stuff will get you the user interface.

Mm-hmm.

Um, now when you want to start thinking about backend functionality, there's, that's where you'll dip into a software developer or systems architect, or maybe,

it lingo,Understanding, being able to speak intelligently about APIs,just what is an api,

right? Security

Exactly. Payroll

kind stuff.

You don't need to know how to make these things, but you do need to be able to describe accurately what you want to occur inside your app.

So just to give you a good example, if you're not familiar with an API, it's not rocket science, you'll be able to figure it out the concept pretty quickly. It's basically how two pieces of software talk to each other. There you go. That's probably actually a good amount of what you need to know there.

So if you want your piece of software that you're building in Bolt to talk to another piece of software, let's say open AI because you wanna build an AI powered app, now, I want to connect to the Open AI API and send the text that I put into my app to OpenAI, and then it's gonna send some text back and I want to display that here, right?

That's it. Like being able to use the real words and phrases to just describe what you want to occur will get you extremely far. And I'd say that's We don't have a time for the whole course on everything I know about

software. I know. We should just come to my podcast and we can just throw a course.

Do you wanna side hustle, Alex?

Yeah, we could absolutely do a deep dive session on what are the core skills, but I'd say that's a good intro to the first skill. Yes. Learn the language.

I feel you're almost telling me the core skill is a, having a curious mind so you can keep digging. Because the biggest fear I have is, you know what?

I don't know what I don't know

a hundred percent. And that kind of, that goes back to my whole philosophy of about all jobs being the same, right? if you have a objective and you're curious about how to solve it, and you're willing to dig in and learn and ask questions. of people, but also very conveniently of ais which know this stuff extremely well.

that's really, at a base level, all you're gonna need to figure out everything that you're going to need to figure out to be successful with making software in 2025, and it'll be even better in 2026.

Are you seeing this as a trend that more people, 'cause you originally were targeting at developers and then you wanted to position to everybody, and are you seeing this as the trend?

Where are we in this chasing chasm trend of. Adopters from everybody. Tools like this.

Yeah. It's interesting question because I think in some sense, everybody's asking that question at a societal level. yeah, are we on this adoption curve where AI is gonna permeate every part of society and be adopted by everyone?

So it, my feeling is that we're still in the relatively early phase. Like we're in the early adopter phase of the curve For sure.

So if you're watching this and you're, you feel like, oh, this started 14 months ago. I'm behind, I mean. Yeah, you are a so, but you're also ahead of 99% of the population that haven't even heard of these tools and don't even, maybe they saw something about AI on the news or something, but the, we're still at the point where the majority of people in society have not actually picked up one of these tools and tried it, which I highly recommend you do because it's so easy to get started and your mind will be blown and then now you know you'll be off to the races from

there.

Yeah. Can I make money off this? These apps vibe coded through Bolt. what more do I need?

Yeah, 100% you can, and the only thing that you need beyond Bolt is a good idea and really, and maybe some distribution. I think what we're really seeing is that the importance of distribution and go to market is becoming a lot more important as it becomes easier to take an idea and turn it into software.

Getting it out there in the hands of users becomes more important. But there's always, there's always some excuse for why you shouldn't try and you can't succeed. The excuse used to be you didn't know how to write software, hey, now it's just, you gotta tell people about it. So that's a heck of a lot easier of a problem for you to go figure out.

you could start a YouTube channel like Angelina's done here. That's a. Becoming a topical expert in the space that you're trying to spread your product into is, you know, tried and true. but yeah, the reality is we're seeing a lot of people making money with Bull. And one of the, one of the really awesome early ways that I've seen that I would, this is where I'd put my money if I wasn't working at Bolt and I needed to start a company with Bolt instead, is, agencies.

So there's so many small and medium businesses in America that could use much better software. As you well know, if you've ever interacted with a small business that has software, looked at a local services business website or tried to make a reservation to have somebody come out and inspect something at your house on through their website, there's a lot of companies that would benefit, from having better software.

They're not gonna hire software development teams. They're busy serving their customers. And they need cheap, fast solutions. And that is what, web development and software development agencies are able to provide with tools like Bolt. so that's a big opportunity that we are leaning into and looking for ways that we can support our agency customers, more.

But, that, that's where I'd start. And then develop your skills, get some cash flow going along the way. Along the way you'll have that idea, where you could really turn it into a product or service and go from there.

I saw a recent episode with DOA C Steven, Bartlett.

he was interviewing chris koerner he's the king of side hustles. He was talking that this is, at the moment one of the hottest small business or freelancer opportunity out there is AI consulting, which is helping SMBs integrate ai.

Completely agree.

And you can use both to do that. All you need to do is, like what you said, distribution and reach out, go to market to get to talk with the small businesses

a hundred percent. And it's never just a little free sauce here. It's never been easier to land a client for a business like this because in five minutes you can have a prototype of whatever it is you want to sell them with their marketing copy off their website, their images off their website, the same sort of color pattern if it's not atrocious.

you can really show instead of tell in the sales calls. So it's never been easier to land a client.

Okay. So this is important to help my audience or anyone who wants to do this, to understand how do you compare both with lovable cursive or cell, and there's also the clot cold and now the Google anti-gravity, right?

It's almost like this market is a pretty good idea. 'cause it seems like a lot of people have entered

Getting crowded.

Yeah, absolutely. So the way I would, position bull is that. We are easier to use than cursor, especially for non-developer users. And we're more powerful and enterprise ready than the other vibe coding tools in the space.

And to just give a quick nugget of, it's what's easy to understand why we're easier to use than cursor we in the brows, you just go to bolt new type text and hit enter. There's nothing to install, there's no scary developer looking interfaces or vs code or local dev servers or GI required or any of that.

but on the side of how we're more powerful and enterprise ready, this is a little shout out to the digital product teams, folks that are on product design or engineering teams at big companies. and this is where I'd love to, to work with you guys as you well know. Calling back to early in the episode, we talked about design systems.

In order for any of this AI generated code to get implemented at the enterprise, it needs to align with your enterprise code requirements, correct. Tech stack security, et cetera. And visually it needs to be built using your company's design system and visual styles. And that is where a lot of these tools are really falling down, and where we are here to help you and your team succeed.

We come in, we help integrate your enterprise code requirements so that your engineering team, instead of being maybe a bit apprehensive about this AI generated code, you're sending their way. Instead they're thankful because it's compliant and it uses the design system and it can actually be used and merged into the enterprise, code base rather than just being a prototype.

How do you ensure accuracy and then the amount of control, the level of hallucination, right? This is a big problem. I've used cursor and then they give me a lot of hallucinated results and I stopped using it.

Yeah. So to be honest, like this is the hardest problem at the cutting edge of what we're working on.

And the way we're solving it is we are actually deploying our engineers into our customers companies to work with them to actually build their. Their code requirements into BOLT for their organization. And we're doing r and d with our customers to get that perfectly dialed in. 'cause the reality is that's not a completely solved problem at this stage in the market.

I highly suspect that by next year we'll have a very product ties solution where it's just like you feed in your design system code here, you feed in your documentation here, you put your enterprise, security and tech stack requirements here, and we will have a fully automated way of handling that.

but yeah, at this point it's us working hand in hand with our customers to ensure that the output that they're getting meets their requirements.

and I also have a question about how does impact organization structure, right? So you mentioned this will enable the PMs and then to iterate through their ideas and have validation.

Before they move this to get, executive buy-ins or move the actual implementation to the engineering team, would that, give them a lot more power than the engineers? And what does that mean? I mostly wanna know how we should think about building a startup. how do we build a team in the future?

'cause the dynamic is changing right? Before I'm thinking maybe I need five people on the product team and 15 people, or at least on the engineer team. Yeah. Maybe I should think this different.

okay. It's a great question and it's one that everyone is asking themselves right now. the book has not been written on this question, but I'll give you my take right now.

Certainly we're seeing the role and. Structure and operating flows of the product design and engineering teams in flux at the moment, and I don't think anyone knows exactly where it will land, but my encouragement to everybody is to lean in and work together to figure out how these tools can improve your processes and flows.

I don't think that by definition they. They disadvantage any member of that product, design, engineering, collaboration team. I think when used properly, they're bringing those teams tighter together. They're reducing the friction of the handoffs between those teams. they're ensuring that the fidelity of the idea that everyone's trying to build together can flow through that iterative cycle between those teams in a higher fidelity way, in a quicker way, and in a more collaborative way.

So time will tell how that affects exactly the structure and the number of people. but this is exactly what we're working with our customers on, is how do you integrate this tool and improve the collaborative workflow for everybody on all three of those teams.

Yeah. there's more to see and definitely maybe I will chat next year again.

I'll ask you, are you seeing more organizational changes. It's affecting everybody's work.

I just started a closing tradition, for the podcast. if you were to recommend one book to a 20-year-old, one to a 40-year-old and one to a 60-year-old, what would they be and why? Okay.

No

question on all three, but whatever you can recommend.

Absolutely. So for the 20-year-old, I would recommend a book that I read a little bit before 20, when I was, I think around 11, if you haven't read it by 20, it's not too late. It's called Rich Dad Poor Dad by Robert Kiyosaki, and I know,

yeah.

Yeah. I think that's a book that you need to read before you graduate college. It's not a technical how to manual on how to make money whatsoever. And if you read it that way, I pray for you because it's not gonna work out well. But it does instill some really important concepts that I think are important to understand before you start your career hm mm

I like the framing of this question. a lot of my books are probably just, you should read them all before you're 20. 'cause I haven't really figured out necessarily the ones you need to read at 40 and 60,

So I figured another one that would be good would, for the 40 year olds would be this book, the Intuitive Parent. it's by my father-in-law, Steven Camarata.

And it's about child rearing and how the most important thing that your, child needs is you and not necessarily, Mozart for babies and all this stuff that You're probably getting bombarded with Yes. When you're a new parent. so that one I thought would be good for 40 year olds.

And then for 60 year olds, I just had to go back to my favorite book of all time, which is Enders Game by Orson Scott Card.

And if you make it to 60 and you haven't read that book, you should go read it. it'll remind you of what it's like to be a young, curious adventurer. So those would be my recommendations.

Awesome. I love those. I can get prepared for all those. I probably should reread the Rich Dad and, poor Dad kind of the book.

maybe I'll have some new takes

perhaps. hey, DM me about it. if

you're

reading it, always happy to chat about real estate and investing

Yeah.

Yeah. Thank you Alex. thank you so much for, joining me with the chat today. I really enjoyed chatting with you and I know that you shared a lot of the insights about how to make, both go viral and went through this hard moment and become the product that you guys are all building towards.

Absolutely. Thanks for having me. thanks for having me and I, really appreciate the invite and look forward to chatting with you again.