V0: Generative UI
vercel.com
vercel.com
It's like v0 but for HTML and TailwindCSS +/- vanilla JavaScript.
It builds UI from Figma as the starting point, but the UI can then be iterated with prompting or made interactive, etc.
It’s based on their work with mitosis so it can already create components in a variety of frameworks.
I’d say you still have to know front end dev to do anything useful with it, but it can be a decent accelerator.
[1] https://www.builder.io/blog/figma-to-code-visual-copilot
Make a similar site to Basecamp, Notion, or something much simpler like an Online calculator? So people can bootstrap their App idea without even being a professional programmer?
Well GameMaker came out in 1999. So maybe like 2002?
once you have the prototype you can then give the AI some of your observability data and spin up a sandbox environment and have it iterate on any code that is slow, have it create 10 new versions of the code and send some mirrored traffic to test it and then deploy the best version.
NextJS and HTMLx are back to server :)
pnpm create next-app --example blog-starter
And the resulting directory is a whopping 338M monster. Wondering what this adds.
anway about v0: this open box to start a project is the AI feature with lowest effort, they are bringing nothing not even a better UI but they have the userbase. Which is good to bring more customers to AI apps.
Not sure why it's showing posted "5 hours ago".
Happy to answer any questions!
Awesome use of LLM. Do you position V0 as helper for developers or designers?
My vercel login is with github, same username as here.
The perfect addition here would be to also enable users to generate React components that visualize each content component (I was able to prototype that already with ChatGPT, but I assume what you're doing is more advanced).
Wrap all of this up into a Git repo with Vercel deployment workflow for GitHub and it should be possible to go from prompt to CMS-driven website in a really short time.
So, my question: Would it be possible to get access to V0? Also, feel free to reach out if you see potential for collaboration.
I actually tried to recreate the functionality with Cody [cody.dev] in VS Code and it actually worked pretty similarly minus the GUI, but I found it a little less restrictive and I could specify the CSS framework.
(But I don't know if this was part of the pool. HN's ways are mysterious.)
What happens when you need to debug this?
> v0 produces code using open-source tools like React, Tailwind CSS, and Shadcn UI.
So you can just debug it as always. I guess this product is for someone that COULD make the app they’re generating, but chooses not to. Rather they want an app, that they then only have to refactor.
Jokes aside, I think it’s an interesting use of generative AI.
In theory, you'd save a bunch of time and then use that extra time to add more polish to the product. But in practice, what you'd actually be doing is reading someone else's (or something else's) code, and people who want to rely on AI are not the same people who have the stomach to do that.
So they're either going to (a) abandon the generated code and write their own, in which case, what was the point to begin with? Or (b) they're going to just say "fuck it" and push the generated code to prod.
Neither of those are good outcomes.
We're embracing copy-and-paste with the product rather than increasing abstraction. The tool provides the HTML or React code for building the generated UI. The way you actually use it, is by copy-pasting it into your app. From that moment on your workflow and specially debugging workflow is exactly as before. The tool just writes the first 100 lines of code to get you started with a great UI.
Maybe it'll push us for the final collapse of the web though. This video is very relevant at this moment: https://youtu.be/ZSRHeXYDLko
1. Tailwind is css but "disconnected" from other parts of the result. So it could be easier to generate.
2. The code that was used for training, often used tailwind and thus the ai can work better with it.
Tailwind is indeed required at the moment. However, we're working on supporting other popular design systems as well as custom ones.
I may be getting far too cynical for this stuff, having seen it rise and disappear over and over, over the last ~40 years.
Bravo!
Edit: Why the downvotes? I don't understand this at all?
What's great about this is that not only this is a next step in AI generated websites, but this is actually shipped software which is much better than the overpromised and overhyped thegrid.io vapourware.
https://developer.mozilla.org/en-US/docs/Web/CSS/CSS_grid_la...
I was referring mostly to GSS: https://gss.github.io/
I think it was an interesting idea that unfortunately didn't really work out. This was before both flex and grid layouts were a thing, so anything new in this space was exciting really. It's a much better css layout world today than it was 8-10 years ago, and particularly grid layout is a very welcome change, but I can't help but think some great ideas have been left by the wayside. Maybe for good reasons, but still.
I hope you understand this is a stupid game to play. We could chase our tails with this all day. Metadiscussion is off-topic.
"make the company logo blue". This is only game-changer if you also think React-PHP (RSC) is game-changer (which was, but to stop React growth and start its demise).
I think it is fair to be skeptical.
Roll a d20 and add your Wisdom modifier.
With disadvantage because 2 ≠ many.