Ilan (00:00)
out... I don't know, this joke... I didn't have the rest of the joke, I was hoping it would come to me.

David Vuong (00:05)
I started the sentence and I didn't know where I was going.

David Vuong (00:11)
Hey everybody, welcome to Prompt and Circumstance. My name is David.

Ilan (00:15)
And I'm Ilan

David Vuong (00:16)
And today we are going to be reviewing Bolt.

All right, our agenda for today is the history of Bolt, a live demo of Bolt,

A deep dive into some of the key functionality. We're look at their strategy. And then finally, we're going to decide on whether or not we are going to recommend Bolt.

Ilan (00:44)
But first, let's hear from one of our sponsors.

We're excited to be sponsored by our friends at Querio. Do you have too many data sources and not enough time or know-how to link them all together so that you can get data informed answers to your product questions? Querio has just the product for you. Their AI agent sits on top of your data stack and allows you to ask natural language questions to get data responses from your various sources. I personally use it.

and it saves me hours per week And just for our listeners, they have a special deal. If you click on the link below let them know that David and Ilan sent you, they'll give you two months free of Querio Try it now.

Ilan (01:32)
before we get to that, David, why don't we talk about a little bit of recent AI news?

David Vuong (01:36)
Yeah. Awesome. so Anthropic, which is the company that makes Claude, they released a new foundation model. in fact, two, they released Claude 4.0 Opus, the larger one and Claude 4.0 Sonnet. And this is really important, especially for the vibe coding community, because those might've been the foundation models that, are used by, these vibe coding tools. So.

3.7 was the previous one. And is the commonly used foundation model for some of the vibe coding so I think we can expect to see an improvement in what these vibe coding tools are able to do.

The system card shows that it's going to do a better job of doing only what you ask for it to do instead of going about and doing all these extra things. I mean, it's never done that for your code, right? It's never gone and changed something completely unrelated to the fix that you wanted it to do, right?

Ilan (02:32)
Yeah, that's right. No, that's never happened.

David Vuong (02:34)
In other news, there is also ⁓ Codex being added to the offering from ⁓ OpenAI. So this is a tool to generate code. This is a coding assistant by OpenAI. And we can see everybody is really hopping on this bandwagon of AI tool assisted development.

Ilan (02:58)
Yeah, it'll be interesting to see because we previously mentioned that Windsurf, which is a ⁓ code assist tool is in talks to be acquired by OpenAI. So they're definitely parallel tracking this in a few different ways. But one thing with Codex that was interesting is

It seems to be optimized around making PRs in GitHub and testing the code. So I've even seen a few flows where people use Claude to write their code and then Codex to do the pull request testing portion.

David Vuong (03:37)
That makes a lot of sense. I heard ones where somebody made, code with Gemini and then tested it with Claude and vice versa. So maybe the software development team in the future is going to be Claude, Gemini and GPT.

Ilan (03:51)
Yeah, I absolutely think so.

David Vuong (03:53)
You know, with, with tools like Codex and Windsurf.

You would think that they actually belong in the same category as Cursor and GitHub Copilot, where the intended user is a developer as opposed to necessarily, Replit or Lovable, where it's more about that citizen developer, right?

Ilan (04:15)
Yeah, absolutely. Speaking of the Citizen Developer, we both got an email from Vercel letting us know that v0 is changing their pricing model to be much, much clearer for the user.

by adding a very unclear token-based pricing scheme on top of their regular monthly charges. Of the four tools that we've reviewed so far, including the one we're gonna do today, they were the only one who did not have some kind of usage-based pricing. So not shocked to see that they added that

The types of tools that they use in the background to process your requests are really expensive and they charge per request. So I can see why they did it, but I guess we'll pour one out for the last of the big Vibecode tools that didn't have usage-based pricing.

David Vuong (05:07)
That's right. Yep. Now they're all priced with a similar structure and they can all race to the bottom for us.

Ilan (05:13)
That's right. It's amazing how this industry went from being novelty to highly commoditized in the span of six months.

David Vuong (05:25)
That's the, that's the AI era for you. All right, let's go to the show.

Ilan (05:28)
That's right.

David Vuong (05:31)
All right, so let's hear a little bit about the background on Bolt, Ilan

Ilan (05:35)
Yeah, absolutely. So Bolt was started by Albert Pai and Eric Simons, and it was actually started as StackBlitz. And the whole idea was it was going to be the IDE for the web.

And after seven years, they were about to shut it all down and Claude 3.5 came out and they're like, Hey, this is pretty good at writing code. Our browser based code development tool might be really useful for running code that's developed in Claude. And they launched Bolt in a tweet. And after five months, they hit 40 million in ARR. They're on track to a hundred million in ARR this year.

They got over a million users in that first five months and that was as of about six weeks ago. So you can only imagine how they've grown since then. It's wild. It's wild. And I mean, the stories are similar between all of these tools, right? I think we just have to get used to in the current day and age companies going from zero to a million in no time flat. But...

David Vuong (06:27)
What a rocket ship.

Ilan (06:45)
Yeah, they have a lot of interesting marketing movements. They do a lot of marketing on X. The founder, especially Eric Simons, is super active on X. And actually, funnily enough, they got tweeted at a little while ago about, hey, if I was the founder of Bolt, what I would do is I would try to break the Guinness World Record for the biggest hackathon ever. And...

You can still sign up this Friday. They're doing the biggest hackathon ever. They're trying to break the Guinness World they're giving out a million dollars in prizes for that.

David Vuong (07:19)
That's exciting. I mean, to be part of not only a Guinness World Record, but also to maybe win some cash as part of that.

Ilan (07:26)
Absolutely. David, should we join?

David Vuong (07:30)
do it. Let's do it. We'll bring all of our little demos and we'll see what we can do.

Ilan (07:36)
So yeah, that's a bit on Bolt. I'd love to jump into a live demo of the tool, show you what it looks like.

David Vuong (07:43)
Let's do it.

Ilan (07:45)
All right, so here we.

David Vuong (07:45)
Ooh, look at that.

Look at that dark IDE. Look at how hackerish that is.

Ilan (07:51)
Exactly, exactly. I have to say that just coming in as a new user, this is probably the cleanest interface out of any of the tools that we have looked at so far. one interesting thing I'll point out here is they offer that you can start a blank app.

Or start with your favorite stack. And this goes to something that you mentioned, I believe in the last episode that most of these tools, they're focused on React native apps. That's something that Claude is particularly good at programming, but in Bolt you could actually choose, Hey, I want to build an angular app. want to build with Remix or with Vite. ⁓ So

as with every other one, you can start to and...

it'll Prompt you to create an account

David Vuong (08:40)
Yeah, very cool. What I also noticed is that they're, you know, unlike the other tools, it doesn't have ⁓ example applications. You know, with some of the other ones, there was a community of, hey, here's some sample projects that others have done. Here's some sample apps to inspire you. But Bolt is only just the Prompt and ⁓ the buttons to start on a particular stack.

Ilan (09:08)
the other hand, you don't get the inspiration from what other people are creating to help guide you. I don't know, do you think there's one way that's better than the other?

David Vuong (09:17)
⁓ I'm not particularly leaning one way or the other. I do appreciate how clean it is. It does have these little prompts underneath like these starter prompts, like build a mobile app with Expo. Start a blog with Astro. ⁓ so those things could be, sufficient inspiration. I'd be very curious to see with the other apps.

what the conversion rate is on people actually clicking through on those community apps to see what they're like.

Ilan (09:48)
Yeah, I remember hearing from v0 that it made a marked difference for them in retention of users because people felt like they were able to get to something great sooner. But yeah, I'd be curious to know as well.

David Vuong (10:05)
Something for Bolt to add to the backlog.

Ilan (10:06)
That's right. ⁓ A couple of things to note, like the other tools, there's a Figma integration and also a GitHub integration. So you can bring in your existing projects and designs and start from there.

I'm going to give it a pretty...

Quick Prompt, you know, one sentence, but I'm going to go off of something that we tried in a previous episode, which was to build a mobile teleprompter for ⁓ podcast notes. And the other tool we tried it in got stuck. I believe that was v0. So let's see what happens here.

David Vuong (10:42)
All right, let's do that. So the interesting thing about Bolt is that right front and center underneath the, the giant hero, ⁓ message of what do you want to build? It includes this subtext of, saying that they can deploy on mobile apps. So that's a really quite the statement. It's a bit of a differentiator. So away we go. ⁓ Ilan's prompted it to say, build me a mobile teleprompter.

for podcast notes and it's gonna go away and think about that.

So it's making a plan it looks like.

Ilan (11:17)
No,

that's right. This is somewhat similar to the experience we saw in v0 where it makes a plan for nine seconds. It gives you a wall of text and then it just starts going and

David Vuong (11:31)
That's

Ilan (11:33)
It's gonna start using the tokens right away that you have in your

David Vuong (11:40)
Yeah, ⁓ there's no pause to say, hey, here's my plan. Are you okay with this before I go away and start building stuff?

Ilan (11:51)
For the impatient PM, maybe this is a feature, not a bug.

David Vuong (11:56)
Yes, that's right. what I, what I'm appreciating is the, the styling, honestly, of, ⁓ of all of these tools, actually, there, there's this kind of trend where the input text box has this gradient, border around it. ⁓ and, ⁓ gradient fills seems to be trending these days. I don't know if you've noticed.

Ilan (12:23)
I

David Vuong (12:25)
but especially

anything to do with AI or anything that's new, they have this gradient fill on the text.

Ilan (12:31)
I have noticed that,

So I'll note that we're still working here. Bolt is not the quickest tool, but it's doing a lot of stuff.

David Vuong (12:44)
One of the things that I noticed is that these kinds of tools, if you don't give them guardrails, they will build a lot of extra stuff. So in one of my experiments, I asked it for a simple micro application where all it needed to do was call some APIs. But because of the nature of the data that I was working with, it felt the need to add a whole authentication user flow.

to it. So in order for me to start to go from the initial page to the page where I can actually do some stuff with the data, I had to go through this authentication workflow. to log into my own app.

Ilan (13:27)
You

they're definitely making a lot of decisions for you and I guess for them it works well because then you spend more tokens with them to fix the things that it decided on your behalf that you might want.

David Vuong (13:49)
⁓ some dark UX patterns.

Ilan (13:52)
Well, we're done here. Bolt has finished building the app and on the screen for those who are listening, we've popped up a little window that says preview on your own mobile device, install Expo Go and scan a QR code. So it does require you to add a separate app in order to test out your mobile.

app experience but I did that before we recorded and so I'm going to show you how this looks.

Ilan (14:22)
But before we get to that, let's hear from one of our sponsors.

Do you have a side hustle? Why not? If the answer is not enough time, worried about validating, or not sure how to take the first step, our friends at Co.lab have just the product for you. With their Validate and Build products, they will work with you side by side to take the effort of validating your product with your market, and then building an MVP and finding your first paying customer. They'll hand off a working product

to you so you can run your side hustle from that point forward. Gain independence and check out their link in the show notes below. Let them know that David and Ilan sent you for $250 off.

Ilan (15:10)
Alright, so what I'm going to show here is the app that I'm seeing on my phone, which is ⁓ it looks like a native mobile app. I have a big button in the middle to create a script. I can enter the title, ⁓ Episode.

Four A New Hope and enter some script which I'm just gonna put in some random characters and

then I'll save it. ⁓ my script doesn't show up to be saved. ⁓ So unfortunately, it didn't work to save my script, but then I have a prompter and I have a settings page. This is a pretty complete mobile app and not bad for a one-shot from a single-sentence Prompt. Very curious to see where this could go if I continued.

David Vuong (15:54)
⁓

Ilan (16:16)
to Prompt this and fix some of the features like being able to save my actual script.

David Vuong (16:21)
you could launch it to the podcasting community.

Ilan (16:24)
That's

right. Maybe I'll do that. So that's that. That's the live demo.

David Vuong (16:32)
Well, I mean, you know, let's, let's

go back even four years and say, Hey, I have this website where you can give a simple one sentence and it's going to make a mostly functioning app for you. Okay. What, kind of magic would that be? Right.

Ilan (16:52)
Yeah, I mean you can go back four months and That's still you know it seems it seemed ⁓ magical then right, so yeah, it's ⁓ it's really incredible with that David let me turn it over to you to take us into a deeper dive

David Vuong (17:07)
Yeah.

All right, so here I am in Bolt and ⁓ as you might have noticed, this is the similar dashboard mock-up app that I tried to make on a previous episode. And I gave it the same Prompt because I was very curious how these ⁓ tools would compare. And in fact, you might notice that I have the other tools up there and we can totally compare. So as I scroll through this,

You'll see that it did a similar thing where it made a lot of different pages. And I actually asked it to deploy it. It did it all right. And here we are. So again, this tool is meant to be something that allows people to mock up a dashboard where they can choose, all right, I want to see a bar chart here, a line chart there, and so forth.

⁓ so if I click on one of these, make a stacked bar chart, Hey, there we go. A line chart, maybe an area chart, a pie chart, gross pie charts. so, ⁓ what's, what's good is that, ⁓ it is able to, ⁓ move things around, although it's a little bit janky. That pie chart was not where my Cursor was at. ⁓ but it is able to move it around. Okay.

Ilan (18:15)
haha

David Vuong (18:31)
which is more than I can say for some of the other tools, it ⁓ cannot resize. And now this is just the first go at it. So I can Prompt it to allow for resizing if I want to, but I'm curious how this compares to others. So if I were to say, change the color scheme, that does not work. That didn't work.

Ilan (18:58)
Hmm.

David Vuong (19:00)
And ⁓ if I come over here to say v0, looks quite similar, right? And if I were to say bar chart, line chart, maybe a stacked bar chart, notice the difference here in logic. So I am able to move this around and I believe, yes, I am able to resize this, although the stacked bar chart looks broken.

Right? So look, at least the functionality is there. Would you like a single bar in your bar chart? That's all right. But look, the functionality is there. And if I change the color palette. wow. Look at that. Let's change this to like a ⁓ red. How about that? Apply. Look at that. Color palette changing works. Let's compare this to Replit. Here we are in Replit.

Very similar layout, very similar design, isn't it? It's almost as if the foundation model they use is identical. How about that?

Ilan (20:02)
Yeah,

that's right David, I wonder ⁓ how that happened.

Well know, LLMs are really good at doing things in the style of somebody else. I wonder if you asked that, hey, could you build me a dashboard mockup builder but in the style of Edgar Allan Poe, how it would do.

David Vuong (20:23)
that'd be something.

Yeah, so color palette working all right over here in Lovable, similar thing, stack bar. Notice some subtle differences though, right? Like here, I actually have some toasts, which is nice.

⁓ And here's the table and so forth. Color palette working okay. So it's interesting how similar the end result is. And even the layout and some of the things there, again, I suspect that that has to do with how they are possibly sharing the same foundation model or like you said, Ilan, that they are inspired by the same things. ⁓

Ilan (21:06)
Mm-hmm.

Interestingly though the As you said the the UI is the same but each one or is similar, but each one prioritized a different aspect of the the Prompt that you gave Lovable for example, you know, mentioned the toast there this is one thing that they brag about they're really UX focused and so

you know, they use the UX best practices in theirs. They focus on getting that done, but maybe not some of the other features that you would think about as a user, as a

David Vuong (21:53)
That's a good maybe the, the, the system Prompt is something that made made that difference. And you know, it's totally okay to share the same foundation model.

Right? Like, ⁓ you know, I'm, I'm in the, heavy equipment industry, ⁓ where, know, our customers would be selling excavators and, and bulldozers, those kinds of things. ⁓ and it, it is the case that some of them have the same engine, the actual engine itself is the same. And so then how do you differentiate? We differentiate around that. Right. And so applying that in a, ⁓ vibe coding tool context, how you differentiate is.

the user interface, it's the system Prompt, that those things now become what your actual IP is.

Ilan (22:45)
That's right. one thing that I wanted to has its own error monitoring. And one of the things that I liked there was it gives you two options. You can attempt to fix or discuss the problem.

David Vuong (23:00)
That's a really good point. And here it is just underneath the Prompt text box where I can switch from discuss mode to build mode. So if I were to discuss this, this app that I've got built, I could ask something like, what would you propose if I were to make this into a mobile app?

So now I wonder whether this would spend tokens. I think it would. speaking of tokens, the cost to build this mockup was about 60,000 tokens, which is not bad, I suppose. I think that depends on, you know, your pricing threshold for tokens. We can dive into that when we get into the pricing.

Ilan (23:53)
is really cool, David. I appreciate you taking us through that and some of those features. Do we want to talk a little bit about the company strategy and how they compete and we could dive into that token pricing that you were talking about a moment ago.

David Vuong (24:08)
Yeah, sure, let's do it.

Ilan (24:11)
So how does Bolt stack up to all of the other tools out there? How do they compete? What is different when you come into Bolt versus the other three tools that we've looked at?

David Vuong (24:26)
I got to say that from an initial user's perspective, there isn't anything that immediately jumps out at me as being tremendously differentiated. Now, having said that, that assumes that the person would have done this kind of comparison across the different tools. If Bolt was the first and my only tool that I've come across, I would be blown away by how great it is.

Ilan (24:45)
That's right.

Absolutely. The things that I would mention that we saw with Bolt that do differentiate it a bit are the mobile experience, the fact that you can create a mobile app. It's not just a responsive app, but a fully mobile native app. That was something that we didn't see in any other tool. So definitely one differentiator if that's something that you're really looking for, but otherwise I agree with you, you know, overall the

look and feel pretty similar to ⁓ some of the others.

David Vuong (25:25)
Yeah.

Ilan (25:25)
I would say

where they are a little different is in terms of their PLG motions. So their pricing scheme, we both noticed is a bit different from some of the other tools on an annual plan. They're only 18 bucks a month, which is about 10 % cheaper than the next cheapest tool.

David Vuong (25:50)
They're undercutting the competition.

Ilan (25:52)
That's right. But a key thing is that they charge you for tokens along the way

David Vuong (25:58)
It's your, it's your

consumable. Yeah, that's right. And you know, my take on their pricing being based on tokens is actually quite positive. I appreciate that. ⁓ they, ⁓ price based on tokens because, ⁓ it does feel a lot more transparent to me now that's because I know what a token means. However, for people who don't know, ⁓ they could, ⁓ look up what a token to word.

exchange rates would be And it's, it's about, ⁓ you know, ⁓ for every four tokens, it's about three words or so approximately. Now it does depend on the word if you want to get technical about it. But nevertheless, because it has that ⁓ very close correlation to number of words, to me, it feels a lot more fair than ⁓ something a lot more abstract, like credits.

which could be changed in their value in what they represent arbitrarily.

Ilan (27:04)
Now are there tariffs on those token exchange rates? ⁓

David Vuong (27:07)
boy, this is not

so much a ⁓ geopolitical podcast now.

Ilan (27:15)
Well, I'm curious to hear David, is there anything else that you think the listeners might find interesting about their strategy? And two, where do think they go from here?

David Vuong (27:28)
I would say that for a small team and in such a short period of time, they have done tremendously well. You know, we think about some of the competition and how they've been around for much longer for double or maybe triple the time that Bolt has been around. And Bolt has almost caught up. If not, they already have. So, ⁓ I think that there's something about how their team has come together that maybe there's some secret sauce there.

Ilan (27:59)
I think you're right, there's something to that. The thing that I see with them is because they existed for quite a long time as StackBlitz, it's the same team who built Bolt as who built StackBlitz. They had about 15 people, now about 20 people in the company at that time. So they already had a strong working culture, working relationships, they knew how to do this. They didn't come together five months ago, six months ago and start building this tool.

from scratch. So I think that probably had a lot to do with that momentum they were able to generate very quickly.

David Vuong (28:38)
I think that's something to think about when product folks like us whine about how we don't have enough developers to get what we need to get done.

Ilan (28:48)
That's right. Well, what do you think David? Is Bolt a buy or a bypass for you?

David Vuong (28:54)
I think given the competitive pricing and the transparent way in which it's priced, this is a buy for me. Its abilities are really on par with the alternatives and the fact that, as you said, it's able to create something for mobile. That's a great bonus.

Ilan (29:13)
100%, I'm totally with you. This one's a buy for me. actually have a couple other product managers who I work with who are using Bolt in their day-to-day

David Vuong (29:24)
Awesome.

Ilan (29:25)
Well with that, David, you mentioned it earlier, the idea came up, we talked about it offline, so what's coming up for us?

David Vuong (29:35)
Let's do a roundup. Let's have a look at these four tools, Bolt Lovable, V0 and Replit. And why don't we have a very thorough competition across the four of them?

Ilan (29:46)
with that, you got something cool to look forward to from us next week. As always, please give us feedback on...

this episode, leave us a comment, send us a DM. We'd love to hear from you. You can find us @pandcpodacst on YouTube or any of the socials, except X where we're @_pandcpodcast . Like and subscribe and we're excited to see you next time.

David Vuong (30:16)
Alright, see you at the next one.