This weekend I built a usable MVP during the Pi Day contest with @boltdotnew and @AnthropicAI.
Used Claude Sonnet 4.6 throughout and the experience was crazy smooth.
End result:
a real business audit app for SMBs and trades that’s already in a usable state.
Still more to build, but I’m pumped with how far it came in one weekend.
https://t.co/OJntOzZeZh
I use supabase on my projects. Used by 10M developers world wide but by my math, I’m not sure they could’ve done it without us. The math doesn’t math lol
Congrats 👏🏻
Supabase is now used by 10 million developers worldwide!
We want to thank the amazing community we have. We could have done it without all of you!
We will keep striving to provide the best developer platform in the world!
I use both for building and I think regardless there’s no real downside with either.
If you’re using something like replit or bolt to build with, I like the results I’ve gotten with Anthropic. My local stack is Hermes > Codex > Grok.
Probably an unpopular opinion but all the frontier models seem to be at a stage where there are few down sides.
Can’t wait to see what you’re cooking up!
@RealProductGirl Both in the right context. Humans are great but I’m finding AI just gets it done when you leverage it for the right task.
Humans are great too but they’re also human so sometimes that’s sucks lol.
@AlexFinn Loops sound like the future of coding but right now now it seems like a great way to burn through tokens. Local model makes even more sense tho.