Has anyone else used v0.dev to build a real MVP? Would love to swap notes

I've been on an interesting journey with AI-assisted development too. For about 1.5 years, I've been building an AI-first document retrieval engine for internal use - never really planning to release it publicly.

Out of curiosity, I decided to test v0.dev to create a homepage for this product. I was genuinely blown away by what it produced! The quality was so impressive that it actually convinced me to release my retrieval engine to the public (launching soon).

In my experience, these AI tools are absolutely production-ready for frontend work.
By the way https://spyk.io this entire homepage, About page and documentation is 90% AI generated :D

I have. Quite good. https://tinygiant-mvp1.vercel.app/

I am building TextBin using v0.dev. Although, it's not something vast and complex, but yeah the frontend is built almost entirely using v0.dev and few sprinkles of MUI here and there.

Hey Shekhar - I tried to code a directory before Christmas – used V0.dev too.

It was pretty fast but it used to rewrite existing code to something completely different, e.g. I wanted to add something to the design and it also changed other things that were quite good before (I didn't see a reason to rewrite them).

Maybe I just didn't use it properly or should prompt better but for me, it was kind of inconsistent.

If the project is for personal use and does not store any sensitive data, I would ship it.

Otherwise, I would require code proofreading first.

Great questions @nxame !

I would not think of v0.div to launch an MVP... But I use it as a first step, to prototype ideas and share with tech.

In my workflow, v0.dev has replaced Figma.

I'm not a professional designer, nor am I a professional developer, so both approaches are challenging to me when I act as a product person.

I find v0.dev very intuitive, you really can express what you have in mind, it's always close to my vision - or I can fix it quickly. But I also feel limited by the tool, and its technical choices, so as I am more comfortable with coding platform I move my v0 project to one of them (basic code download, and Github upload), where I complete my work.

I recorded a video telling how I "vibecoded" a prototype feature (here), but indeed it's not production-oriented.

I built my base for my SaaS using v0.dev but not the MVP i then shifted to Cursor.ai

Hi @nxame
I've used v0.dev few times, primarily for front-end development or quick landing pages. I found it quite useful for that part of the process. However I typically create most of my app code in either VS Code using tools like Cline or if that gets too expensive I switch to Cursor. Before this workflow I'd typically use Figma and/or Photoshop but have almost fully transitioned to these newer tools now. For quick POC's I use Replit but it's become very powerful so I'm able to get much further with it now. I haven't explored v0.dev for a full application and not much at all in the last while, so it might have become more powerful also. My preference for handling front and back end separately might be more about my personal workflow than the tool's capabilities. Curious to hear if anyone has noticed significant updates or improvements recently with v0.dev. Like you I'm very impressed with what @andres_vlaeminck was able to build with it for his https://spyk.io site.

I wish @Vercel team could add where I can deploy the v0.dev code to a dev or staging environment not production directly and sync my code to a remote git repository!

i have just did a lauch : https://www.producthunt.com/posts/supplementfaqs

Its been over a month that I have been using AI tools for web development.

I have used both IDE based AI tools - Codeium/WindSurf, Cursor and Web based AI tools - Replit, bolt.new, Lovable.dev, V0.dev, and latest Firebase Studio.

Out of all these tools, I recommend Codieum if you are looking for IDE based, or Replit for Web based development.

Replit's Agent is the game changer in terms of speed.

I was able to deploy a simple website to a custom domain under 11 hours using replit (6hrs for dev, 5 hrs for deployment), i faced some issues to deploy it on Vercel first as the developed code was following client/server architecture within the same app. I had to hard code the data in the clent app to deploy it. Here is the link goldomains.store

Later, found that this backend integration to the client app can be done easily using tools like Appwrite (Backend-as-a-service), supabase, etc.

Bolt.new is supporting Supabase straight away along with Payment gateway system Stripe integration. Making it easy for early developers to launch their own small e-commerce apps.

I have been using Cursor lately, and it's context awareness is awesome, but I couldn't find some basic shortcuts that were available in VSCode by default.

Really looking forward to see what these AI tools can do to the prospects of Web Developers, especially Frontend UI developers.

Hey! 👋 I’m in a similar boat — been building for years, and recently gave v0.dev a shot for a new project of mine: ai-world.ai. It’s a site focused on AI news and tools, and I paired v0.dev with n8n for automating data scraping and keeping the content fresh.

Like you, I was honestly surprised by how far I could get with v0. It wasn’t just a quick prototype — it gave me a solid base I could iterate on fast. The real win for me was the speed of UI development combined with how easily I could wire up backend logic and automation.

That said, I’ve started hitting some bumps now that the codebase has grown — v0 sometimes throws errors like "Your message or the text content of your attachment(s) exceeds the maximum length", which makes it tricky to keep evolving the project through the UI. Definitely something to keep in mind if you’re scaling past MVP.

Still, I’m super impressed overall. Curious to hear how others are navigating these kinds of limits or extending v0 projects as they grow. 🚀

I’m just like you—after 24 years of traditional programming, I’ve started exploring vibe coding, and it’s been a lot of fun. V0 has been great! I built careerleap.app using it, spent the last two months learning it, and I’d be happy to share everything I’ve picked up. Here is a quick video I created yesterday to show I how I am using these new tools: https://youtu.be/j5n-Z-DZUVU

Yes, I've used to build an MVP! Let’s connect and share insights Piso Wifi curious how your experience compares to mine.

AI coders can definitely save you time. And if you're more of a product developer than a product coder, tools like v0.dev or lovable.dev are the wrapper to a skill not present in all creative minds.

However, these AI rings of coding power won't work without proper intention. That means preparing your prompts in detail (and by detail I mean in d-e-t-a-i-l), and it won't hurt adding more detail to it, and some more context, and pairing it up with other AIs to fix bugs, and having a little know-how to set up integrations, use SQL, or at least update the code from GitHub to save on some tokens.

Possibly, vibe-coding will soon be contested by technical-vibe-coding, a style where we support our natural language prompts with technical documentation and structured references to help the AI better understand and implement what we want.

Would you actually ship something production-ready this way?

Exhibit A: https://www.producthunt.com/posts/forestalk/

yes I made many apps, but the best was this progressive web app - https://repspaced.vercel.app/

should I launch it?

I've built both popadex and popper using the VS Code Agent with Claude 3.7 extensively. A combination of Agent + Rails 8 and Kamal means really fast development + deployment.

i worked with it for 2 weeks building a dating app but experienced so many glitches and bugs and everytime i wanted to add a detail or feature the entire app would crash and be lost, then i would go crazy and start over. I was so hopeful for this but i think its not quite ready.