Mickey.
By the end of this episode, what are we going to learn?
We're going to know what's the best, what's the worst, what to use, what not to use, and this is just our non-professional, professional opinion.
Okay, so this is the definitive guide to vibe coding apps.
Right?
Yes, this is the ultimate tier list and if this offends you, i apologize in advance.
Okay, any disclosures?
Are you involved in any of these companies?
Way to the end.
I'm actually, no, I'm not an investor in any.
I do have a bias, but I'll explain later the bias as we go on.
Sounds good.
I invested in Bolt in the last round.
Actually, funny story.
I had the opportunity to invest in Lovable.
We both did.
Yeah, but they hit me at a time where I was like, I never had done.
I still haven't, but I was like, I don't know what this is.
Vive coding.
And then, yeah, they raised what?
Some hundred billion at a $6 billion, $2 billion valuation.
Crazy.
Yeah.
Yeah, you win some, you lose some.
All right, so let's get into it.
Okay, so... Lovable, V0, Bolt, Cloud Code, Cursor, Codex, Windsurf, Vibe Code, Rourke.
Let's just hit right off the hip.
What's D tier?
I mean...
I see you touching Windsurf.
Okay.
So I don't think Windsurf is D tier in terms of like tech, but with what happened with the team and like the founder, sort of dipping, and then Devin came scooped them up,
It's one of those where like, I'm sure the tech's good, but like, is there any trust anymore?
And I know two guys who work at Windsurf.
Awesome dudes, but I probably would never use Windsurf ever again.
And respectfully, in the kindest of fashion, I would put it D tier.
By the way, all these tools are amazing.
Yeah, they're great.
They do, you know, like, again, this is our non-professional professional opinion, right?
I just personally wouldn't use Windsurf simply for the fact that, like again you know, you're sort of trusting the team behind said tool you're building with right.
And when I mean you're a founder, like imagine just upping and leaving, like the type of confidence that instills with your users, investors and all that type of stuff.
So yeah, I would respectfully put windsurf at D tier.
It's also very technical, right?
Yes.
Yes.
So, and that's one thing too, right?
We have this chart here and And I would put windsurf at a very technical like you would have to be a technical person to get the best out of it.
Now you have people who like Riley and others who are, you know, using these technical tools.
But if you speak to the team at windsurf or even cursor, they'll tell you yeah, we're building for developers, right?
Their focus is developers.
I actually spoke to someone who works at cursor and they I told him, like you know, a lot of like non-technical people are using your tool.
So is that like changing how you guys build?
And like, is that like sort of, you know, affecting your decisions?
And he straight up was like, no, he's like, we're building for developers.
He's like, it's cool to see non-developers use it, but the main demographic is developers.
And from a team perspective and a business perspective, it makes sense, right?
So you could do it, but I just want to use Windsurf.
So I put it D tier.
Okay, so you're putting a D tier just because of the sort of how it played out.
Yeah, like realistically, I would maybe, it would be between an A and a B for me.
Right.
Because I used to, I used Winster for a bit.
It was pretty cool.
And if based on how it was back then, if I were to rate it, it'd be between an A and a B. Okay.
So, but with all the business stuff, putting it.
Yeah, I'm putting it in a D just because it's pretty complicated.
Okay.
Yeah.
Well, then let's talk about another complicated tool, Cursor.
Okay.
I mean, to me, okay, I'll do this one.
To me, I'm going to give it, it's either A tier or S tier for me.
Well, so first let's define it on the chart.
It's definitely like you have to be technical to use it.
Right.
Like being able to even open it up and fork a repo and PM run and PM install.
Like these are like I don't expect a non-technical person to use it.
But, you know, if there's a will, there's a way.
And I've seen people use it.
So I agree.
If you are a vibe coder and non-technical person, you're going to experience pain.
True, but what I will say about Cursor versus Windsurf, there's a lot more tutorials on Cursor.
And a bigger community around Cursor.
Fair.
And, and, and I'll, I'll give you an example in like the web dev world.
Like there are people who are pro next JS and anti next JS.
And one of the biggest pros to next JS you know people can argue technicalities is the community right?
Especially even with react, like let's go one level down react.
There are people who love react, to hate react, and And the reason React's won is because the community is so huge, right?
So there's a lot of support.
There's a lot of documentation.
There are a lot of external packages.
That's why you go to any of these tools and you ask it, make me a website.
It's going to use React, right?
Even the LLMs have picked it, right?
So I think that matters.
And for the audience sake, I'll give Cursor an A tier.
I can agree with that.
But to me, this is like S++.
Okay, cool.
Sam Altman, the co-founder of OpenAI, just said that it is the era of the idea guy, and he is not wrong.
I think that right now is an incredible time to be building a startup.
And if you listen to this podcast, chances are you think so, too.
Now I think that you can look at trends to basically figure out what are the startup ideas you should be building.
So that's exactly why I built IdeaBrowser.com.
Every single day you're going to get a free startup idea in your inbox, and it's all backed by high-quality data trends.
How we do it, people always ask.
We use AI agents to go and search.
What are people looking for and what are they screaming for in terms of products that you should be building?
And then we hand it on a silver platter for you to go check out.
We do have a few paid plans that take it to the next level.
Give you more ideas, give you more AI agents and more almost like a chat GBT for ideas with it.
But you can start for free, ideabrowser.com.
And if you're listening to this, I highly recommend it.
What's next?
Lovable.
Lovable Cloud, Lovable AI.
So I will- So explain what that is, because not everyone-.
Yeah, so basically the pain point with Lovable and Bolt was really the integrations, right?
You would need a backend and it was really difficult, right?
You had to sign up and then you had to connect and copy API keys.
Again, these are trivial things to a developer.
To a non-developer, this is a headache.
And what they've done from my understanding is they've abstracted all this stuff away.
So the moment you boot up a Lovable project-
I'm assuming they call APIs and they fetch all that information for you, which makes it easier.
But I still have found people running to the cloud codes and the codexes and the cursors.
So I will give lovable a B. How do you feel?
Wow.
Um, I I understand why you're giving it a B.
You know cause you look at it from an engineering perspective and you're probably like well, if I'm going to go and create a product, you know yeah, maybe lovable is something I can get.
You know uh, get going a pretty page.
But, um, I mean, from a prototyping perspective, if you, you know, I'll give you this.
I think V0 is better than lovable.
Yeah.
Why do you say that?
Vercel's marketplace, like their integrations, it gives you a plethora of options, right?
So with Lovable, for example, they've sort of picked what backend you're going to use.
It's super based, right?
And there's pros and cons to that.
Like depending on the type of apps you're going to build, there's pros and cons to it.
With Vercel, you have all different types of what you might call backends.
And here's where the bias comes in.
Convix is going to be on the Vercel marketplace soon.
And I think Convix is the best backend.
And the reason why there's bias is because I work there now.
Just full disclosure.
Because people are going to be like, oh, he's talking about it because he works there.
Well, I started working there because I liked the product.
Homie didn't look at the camera once until he started saying convex.
Said the word convex and he was like, convex is the number one.
I'm PR trained.
But I personally think, and I've for example, I had my wedding not too long ago and I tried using all the AI tools to build like our RSVP site.
And at some point, all of them bricked except V0.
So even if I were just to like vibe code and just go through it, I personally think out of the V zero bolt and lovable, I would pick V zero.
There's also templates and components that you can reuse with V zero.
Yes.
Yes.
And just like the integration with Vercel, it's one of those things too where I know this probably doesn't matter to a tier list, but I also sort of think about like, the future of the, like the tool we're using.
And it's like, it seems like Vercel and V0 are going to be very tight knit.
And that's like a bet I'm willing to make if I'm building a product that I want users to use.
So are you giving V0 S tier?
Is that what I'm hearing?
If I was a Vibe, Vibe coder like, let's say, I did not know how to write code I would maybe right there.
I don't think I'd give it, like can we do an in-between?
Okay, so I'll do it in in-between, and this is definitely non uh technical friendly.
Lovable, of course, is here as well.
You sort of disagreed with lovable being a b tier, so we're no no, do you think it's a?
I can give you like an a between b.
That's what i think it really is.
Okay, i'll give you an a between b.
That's what i think it is.
Bolt i i, i would put them on par with lovable yeah okay yeah, fair see, not a biased investor.
I respect that um, Rourke.
I personally haven't used Rourke, so I will trust your judgment.
I know they do mobile apps right.
So there's this new class of vibe coding apps for literal mobile apps and that's vibe code app.
Uh, by Riley, our friend Riley yeah, and Rourke um, I.
Haven't played with both of them extensively, but played with them enough to know that vibe code app is.
It really is the lovable for mobile apps.
I find the.
Oh, actually there's also anything.
Yes, i have seen anything.
I maybe did one prompt and then i stopped, just because i don't really do a lot of mobile app stuff.
But i i have realized, and even um at convex, there's a lot of like these mobile app vibe coding platforms that are popping up.
Yeah, this seems to be the and for and for a reason.
Right like i think that there's a lot of opportunity to to make money building apps for the.
So you think mobile app like if I wanted to launch a consumer product you think mobile is better than web?
I don't think it's better.
I think it depends what you're building.
I just think that right now there's...
There's just a lot of you know.
You're seeing the Cal AIs of the world making millions of dollars a month with relatively simple apps with AI baked in.
So I think that there's going to be a lot of people who want to build something similar in different niches.
And that's why I think that right now I'm actually coming up with a lot of mobile app ideas.
Really?
Yeah.
You see, maybe I'm just thinking too like I'm too much in my developer mind.
But to me, it feels like downloading an app feels like too much commitment for me.
Like I would really need a reason to download an app.
But it seems like people are just downloading apps like they'll watch a TikTok video and they're like yeah, I want to check this out.
So am I wrong or is like why the push for apps?
Because everyone's doing an app now, a consumer app.
Um, I think, uh, I'm the same way, to be honest.
Like it to me just feels like a lot of commitment.
Yeah, it is a commitment.
It is a commitment.
But like, if you are suffering from back pain and you're on TikTok and you searched, you know back pain for tall people and then you see this ad, that is just hey, are you tall?
And so as a matter of fact, I'm six foot three.
You know well, we have this AI that helps you like reduce back pain by 35.
In this, you know, we become your back pain coach right.
You know, do you want to download this?
And it gets me at that moment.
I might, I might, I might download it.
And I find that in like observing my wife has helped me realize social trends.
People use TikTok like a search engine.
Yes.
So that, that makes sense.
So where do we put vibe code, Rourke, anything, if you've used anything?
I think we can put them.
Right below lovable and bolt.
So B tier solid.
B tier solid.
I think that you can use any of those and, you know, maybe one of them is slightly better on the day or slightly worse on the day.
But I think that they, though what's going to happen, I think is those are, it's so relatively new.
Like I think Rourke launched like a couple months ago.
I also remembered another one.
My boy Seth is a zero, which is a mobile app one as well.
So yeah, I think like to your point.
I haven't tested any of them, want to give a definitive opinion, but yeah, They're all fairly new and it's exciting stuff and I'm sure they have free tiers, so max it out.
I think they're going to get better, too, over time, right?
Do you know, are they all built on Expo?
Yes.
From my understanding, I know Rourke is...
I'm pretty sure they are.
It would be foolish for them to not be, because you would want your app to be on both Android and App Store.
And if you're doing a Swift app, then you're going to have to do the Google Play version.
And that's two different code bases.
So it would make sense for them to use Expo.
And again, the ecosystem is growing.
React is growing.
There's a lot of money behind it.
So I would think that they're all built on Expo.
Cool.
So that, and here's the funny part.
All these guys are probably using the same model underneath the hood.
It's either going to be Claude or it's going to be Codex, which is gaining a lot of hype.
I'm interested to hear what you think about Codex.
I mean, you're the front end developer.
A little higher.
A little higher, yes.
So I seen a lot of hype with GPT-5's Codex model and I used it in Cursor. it was okay.
Like it was pretty good, but I still think clots on it.
Four or five is the best coding model.
Uh, right now.
So I would put codecs like right below cursor, but above lovable and bolt.
Hmm.
And for non-technical people, should they be using Codex?
Yes, Codex does have a web version.
I would put it somewhere around here.
And to your point like a cursor there are a lot of videos on Codex, because they seem to have gone both developer and non-developer attention.
You can use OpenAI's web-based cloud platform to run Codex, so I would put it somewhere around here.
I still think there's a little feat.
There's a little difficulty to it if you're non-technical, but where there's a will, there's a way.
My take on Codex is it's a little worse than Cursor, but it gains the most improved award.
Yes, 100% because it sucked in the beginning.
It really did.
And like now people are comparing it to Claude, like Claude Coates.
So they've definitely, Sam's cooking.
Sam's cooking and I wouldn't bet against them.
No, no.
They have trillions of dollars or whatever, but billions.
Well, it's important, right?
Because if you're going to pick something.
100%, right?
You don't want to be windsurfed.
Yeah.
I'm not mean, I swear.
I'm actually a nice guy.
I couldn't help it.
Claude Code.
Okay.
All right.
What do you think?
I think Cursor's king, man.
Wow.
Claude Code got nerfed.
Claude Code recently has got nerfed.
Yes.
Like if this was four weeks ago, it'd be like this.
Like Claude Code would be above S tier.
But like now, like, I don't know what's happening.
Like I rather like, it'll use the same model, CloudSonic 4.5.
I'll get better results with it on cursor than with on Cloud Code.
And the reason being is a lot of people think like oh, if they're using the same model, it's the same output.
But no, really the difference maker is like yes, everyone's using the same model.
But then you build the agent on top and basically what that means is like when you tell um, when you prompt hey, fix this error right, the llm doesn't have the power to read a file.
You have to give it a tool that gives it the ability to read a file, and then it reads that code and then it gives you the output, and now you have to give it a tool to write that code, right.
So it seems to me cursor's agent, which is built on top of cloud code that gives it the ability to read write, edit and all that type of stuff, might actually be better than cloud, cloud code, cloud code, right.
So i i think this is going to be a tug of war, right?
Like where cloud code will be better at one point, then cursor, then cloud code, then cursor, right.
But as a non-technical person um, If I would be frank, if I was a non-technical person, Cursor would be here, Cloud Code would be here, and this would be the tier list.
We're missing one company here.
Who?
Replit.
Are you an investor in Replit?
No, I'm not an investor in Replit.
Okay, I'm going to ask a serious question.
I don't know anyone who's used Replit, but they're winning.
I'm not even trying to be funny.
I actually don't know a single soul who's used Replit.
The one time I used Replit was back when...
I like, this is like three years ago, four years ago.
Yeah.
And I needed to run some sort of like file, some Python script.
Yeah.
So who uses replica?
I mean, I know two people who use.
Okay.
So they're winning then.
Okay.
Makes sense.
Yeah.
I don't know where I'd put Replit.
I've never built anything fully with it.
People seem to love it.
They raised a bajillion dollars as well.
So I'll trust your discretion.
I mean, from what I understand, they have two modes with Replit.
So they have like your vibe code mode and then they have agent mode.
So you prompt it and it takes like 15 or 20 minutes and it goes and builds your app instead of one minute or two minutes, which is cool.
So they built their own agent.
Are they using their own model?
Are they developing their own models?
I'm not sure if they're developing their own models.
It's hard to say.
But I have seen my friend Billy Howell actually talks a lot about Replit and I've seen some of the output on it and it's pretty impressive.
So you're giving it an a, I'm giving it, I'm putting it like lovable bolt.
Yeah.
So, okay.
So this is fair.
So we have V zero close to, in my opinion, maybe a little in the eight year.
But like if I was a vibe coder, I'd go for V zero.
Codex, lovable bolt.
Actually, can I add one more?
Yeah.
Have you heard of chef?
No.
Chef was actually built by Convex.
But what's funny is the reason Chef was built.
And we built Chef was to show vibe coding platforms that if you use Convex as your backend, it's better.
So we didn't build it to compete.
One or two people worked on Chef and it's actually pretty good.
Like off rip you'll get, and I won't, cause people will be like, oh, you're being biased.
So I won't even show, you could check it out.
It's free.
So chef isn't meant to compete, but it was more so of a showcase for tools to use convex as their back end.
And a lot of the new tools like you have fly, you have bloom, like a lot of the new vibe coding platforms are all built on convex because they saw this and they're like oh, and we open source the repo to anyone can fork and build a vibe coding platform and i non-bias, would put chef Along these lines.
The only thing is again, we're not trying to compete.
So at some point, these tools will get better.
Like, for example, I don't think there's a way for you to assign a domain on Chef.
Right.
So wasn't meant to impress users, but more so developers who are building these tools.
But yeah, I think I'm pretty happy with this list.
Okay, so just to wrap this up, what advice do you have for non-technical people and technical people if they're trying to choose a vibe coding platform?
Yeah.
So if you're technical, then man, it's like cursor or cloud code.
Like there's, there's no other option in my opinion.
And that's simply because you're probably going to be steering the wheel anyway.
So you're not going to be vibe vibe coding.
I know there's like a open code by my uncle Dax, uh, I haven't used it, sorry Onk, I should.
I hear a lot of developers talk highly of it.
So like you're probably using one of these tools and you're probably steering the ship, right?
So no advice to you, just do what you're supposed to do.
Now to non-technical people.
I think there has to be I've noticed after last couple months with all the new tools and all that stuff happening, there has to be a mindset shift.
And the mindset shift is a lot of people get frustrated if in five prompts they don't build something.
And it's like respectfully, the audacity to think you're going to build software that's fully functional, that people are going to use and care about.
Borderline pay you money.
I don't want to crash out.
And you think you could do it in five, six prompts when, you know, it takes time.
It takes planning.
It takes testing.
There's alpha, there's beta.
Then there's a final version.
And then like, software is art and you know, thank the lord, that there are tools that allow the non-technical person to build stuff.
But we need audacity, we we need to treat it with care.
Um, your first version doesn't work.
You used, you know, you used lovable and it broke.
Okay, so what?
You use another tool and try again, right?
I think there needs to be a shift in understanding that software is difficult.
Right, websites are easy, you know, landing pages are easy, but software is difficult, especially once people actually pay for right.
So when we enter this mindset, we're like okay, i'm going to build something, it's going to be awesome, it's going to take me time because i'm going to make it awesome, then we'll be okay.
But i'll see people like yeah, i prompted 10 times and it broke.
Welcome to software development.
So that's the one thing I would see.
Like, there are a lot of cool tutorials.
I mean, you bring up a lot of awesome people that show people like the mindset, you know, planning.
Cursor has this one thing called plan mode, which is pretty awesome.
And all that stuff is great.
And there are even tools that allow you to plan now.
And, you know, ChadGBT5 is great if you want to like just go back and forth and brainstorm.
It's more about mindset.
Now a lot of non-technical people fail to realize that it really takes time to build something awesome.
It's like building a business, right?
Or if you're a gym bro, it's like going to the gym.
You know, great things take time.
They take patient, take calculation.
There's mistakes.
Mistakes lead to experience.
Experience leads to you being a better builder.
So take time and enjoy the ride.
There you go.
Well said.
I want to say one last thing to end.
My recommendation to people is actually to follow all the founders of these businesses.
So V0 Guillermo Codex, Sam Altman, Lovable Anton Bolt is Eric Simons, Replit Amjad, et cetera, et cetera.
Follow them.
And then see who you believe in.
If you really connect with Eric, if you really connect with Amjad, if you really connect with Sam Altman, then pick that platform.
Facts.
Because you are betting on someone.
You're betting on someone.
You're betting on a company, right?
And this is why, again, people who bet on Windsurf got Windsurfed.
Yeah.
God bless him.
All right, that's the episode.
Mickey, thanks for coming on.
Greg, I appreciate you, man.
It's been a fun year.
It's been a fun year.
You've had a busy year.
I'm going to include links to follow you.
I know you're getting to 100,000 subscribers on YouTube.
We're trying, we're trying.
We'll include that there and see you next time.
Appreciate you, bro.
Thank you again.
Later.
