Bubble.io vs AI Coding (bolt.new, V0, Cursor AI, Replit, Claude AI)

sem · 2025-03-06T08:53:45.964Z

What happens if you feed this prompt to bubble AI….?

Important to understand that you need to split it. People think it is easy but after 6 months of only doing production ready apps with only AI we have learnt a thing or two. You have to put in the work and you have to understand at a basic level how code works with all of these tools and AI in general. We even learn every single day and adjust very carefully our prompts.

Should we have 6-12 months of time, I am pretty sure that we can rebuild Bubble. It are amazing times!

bubble.trouble · 2025-03-06T14:06:36.506Z

For anyone who stumbles across this post, please consider that it would be wise not to let a few persons opinion deter you from adding a new tool to your tool belt.

With Bubble you have a screwdriver with AI (particularly bolt.new), you have a high-powered drill. We all still own and use screwdrivers, but most of us prefer a drill for all but the most basic of tasks (and by the way, both tools would be fairly useless if you were working with nails).

Anyone who isn’t using bolt.new should be and I am a huge fan of Bubble, long time user and plugin builder. As much as I love Bubble , the pricing model is wrong and the proprietary platform is problematic (Not being able to add collaborators is an unnecessary and likely harmful “premium” feature - #5 by bubble.trouble).

bolt.new on the other hand is open source and uses open source platforms like supabase and expo. Once we build the code on the platform we can take it anywhere we want immediately including downloading it and running it locally if needed.

It already has support for mobile app (through Expo), integrated database support (through supabase) and one-click deployments like Bubble (through netlify).

It’s not that Bubble isn’t good it’s just that bolt.new (and I’m sure there will be other AI tools in the future) are 1000x better. As has already been said, you have to learn the tool to use it well and it’s frustrating in the beginning, just as even though we may not remember well, Bubble also had it’s own frustrations in the beginning.

Don’t take my word for it though, go see for yourself → https://bolt.new Note: You must use Chrome for your browser with bolt or the web container will not run properly!

(P.S. You will run into errors, don’t give up. Do some research on how to fix them just like you would if you ran into an issue on Bubble. AI isn’t perfect it still needs intelligent human interactions)

fede.bubble · 2025-03-06T18:55:51.790Z

sem:

Important to understand that you need to split it.

Yep, that’s what I’m testing with now. I thought maybe Replit’s new v2 agent would handle large context windows better but maybe not yet in the early release version.

sem · 2025-03-06T19:16:43.549Z

The problem is that you do not know in detail how replit and others handle context. There are many many subtle details that you have to get right to get a complex app working. With the correct questions AI can perfectly do it but you need to structure it meticulously

boston85719 · 2025-03-07T07:06:14.651Z

AI code builders are not replacing Bubble for non-technical people, but they likely will replace bubble for experienced traditional developers as they already know the code and can spot check easily on their own, plus throw it into all the other useful tools they already use for debugging their code more easily.

For non-technical users, Bubble will remain the go to tool, not just because of a need to spot check errors on AI code generators, but also because it is highly likely the non-technical user will be clueless as to how to apply logic properly so to as avoid other issues that are not necessarily mistakes within the code, such as timing issues.

It is also extremely frustrating to be apparently ‘speaking’ to an intelligent ‘thing’ that doesn’t actually follow your instructions every time.

manipaje · 2025-03-07T08:16:08.531Z

From my experience, the direction @nocodeventure has taken seems like a winning path for some projects. You’re combining the best of nocode with that of code. However you’re making the code, you’re basically adding a little complexity for a lot of value. There’s no perfect stack for any project, because they all differ so wildly in their needs, requirements, lifespan, cost, usage, etc. It’s awesome to see more tools, to get more projects off the ground and more people with mixed abilities to get their ideas out there.

I’m still reluctant to pick up platforms like bolt as, from my perspective, I see a lot of social shilling for it.

randomanon · 2025-03-07T08:50:15.812Z

Like, before claiming to build your production app with AI, first build a single plugin with AI. Imagine not knowing how to ride a bicycle and thinking you can solve this by skipping the training wheels (plugins) and moving to motorcross.

joemanier · 2025-03-11T02:30:00.811Z

My bad if late to the party on this and it has already been asked but @fede.bubble, @josh, @kate.mcnally my understanding of what is going on under the hood is that Bubble is translating our visual elements from the editor (text, button, input, image etc) into code, obviously.

Why not give us the ability to do that the other way around?

Why not build us out an area where we start with an AI like Claude Sonnet 3.7 to build us out a UI in code that follows rules to keep it compatible with how Bubble’s visual elements are coded? Then. when we’ve got it just how we want it we say, “looks good, please put it into visual elements” and Bubble does the translation the opposite way from code to visual elements?

I believe it’s much harder than it sounds of course, but, having done a 24 hour challenge to build an AI logo app on my channel recently (shameless plug for No Code Academy), it reminded me how much I love to build in Bubble… except for the UI. Actually I like building UI’s okay but it’s so time consuming compared to the one hour it takes me now to build the 15 UI screens for an app using an AI tool.

It would 2-3x the speed at which apps can be built on Bubble. Cheers.

fede.bubble · 2025-03-11T13:57:22.912Z

joemanier:

Why not build us out an area where we start with an AI like Claude Sonnet 3.7 to build us out a UI in code that follows rules to keep it compatible with how Bubble’s visual elements are coded?

The team is actually working on having this up and running. They’ll be sharing more in the next 2 weeks

joemanier · 2025-03-11T16:56:35.983Z

Sick! Amazing news, cannot wait!

fede.bubble · 2025-03-11T17:08:46.332Z

btw I’m assuming you’ve tried Bubble’s AI Features | Bubble already?

ramzizi · 2025-03-11T20:12:46.094Z

Anyone see the email from BuildCamp with Gregory saying he’s leaving Bubble for Cursor AI?

I’m quite happy with Bubble. I’d rather pay Bubble and let them worry about the servers, backend security, etc.

recker249 · 2025-03-11T21:01:30.928Z

I’ve seen it, I was rather shocked. So is he pivoting away from no-code into programming? Don’t you need some level of programming knowledge to be proficient in Cursor/Windsurf? Even if a prompt generated the code, how can someone such as myself who doesn’t know how to code make edits to the app?

ramzizi · 2025-03-11T21:14:51.854Z

Yes he has a background in traditional coding

randomanon · 2025-03-12T03:33:59.252Z

Very interesting.

boston85719 · 2025-03-12T03:56:47.764Z

ramzizi:

BuildCamp with Gregory saying he’s leaving Bubble for Cursor AI?

No, but doesn’t surprise me that people are leaving bubble for another platform. Doesn’t mean they know something others don’t, it just means they have decided it is best for them. I could imagine there may be more behind the decision than meets the eye.

ramzizi:

I’m quite happy with Bubble. I’d rather pay Bubble and let them worry about the servers, backend security, etc.

I’d be happier if they shored up some things, but I personally don’t have plans to leave Bubble for some AI app builder…soon enough Bubble will figure out what is the key that unlocks their platform (some of us already know what that is).

CodePerson · 2025-03-24T17:45:15.630Z

Gregory at BuildCamp has me freaking out about all of this. I’ve also had several developers tell me Bubble is in real trouble. 6 months ago AI tools were just cute ideas. 3 months ago AI tools began to improve. Today, there is no way you can ignore Vibe Coding. Is Bubble going to add Vibe Coding to the platform? I’ve spent years learning and improving my Bubble skills. The thought or idea of leaving is painful. ARRRGH!

Cheers

hi.luisacosta · 2025-03-24T19:36:06.354Z

The reality is that at some point you will just prompt AI to build/fix applications - whether it’s Cursor, Bolt or Bubble doesn’t matter. Who knows the timeline… I would guess >1 year but < 10 years?

It all depends how well Bubble implements this new expected user experience.

fede.bubble · 2025-03-24T21:22:40.567Z

hi.luisacosta:

The reality is that at some point you will just prompt AI to build/fix applications

this reminds me of of the head of engineering at Figma talking during their summer conference last year. He said AI chat agents should be thought of as another input medium. And as all inputs, there are pros and cons.
For example: you could dictate commands to your computer, but a mouse click is faster and ultimately more convenient.

I don’t think apps will ever be built solely by AI because at some point it just becomes cumbersome trying to chat with an AI on something you could’ve done on your own faster. AI agents are definitely strong with beginners though.

hi.luisacosta · 2025-03-24T21:45:56.697Z

fede.bubble:

I don’t think apps will ever be built solely by AI because at some point it just becomes cumbersome trying to chat with an AI on something you could’ve done on your own faster.

Eventually the UX will dynamically display actions, buttons, etc for the user to take actions and advance via dynamic prompting. Whoever doesn’t do this, will certainly get left behind too.