We let an AI learn taste from behavior, not prompts — curious if this is useful or stupid

Hey PH 👋

I’m launching TasteOS this week, and I want to pressure-test the idea, not hype the product.

Most AI food apps work like this:

prompt → generate recipe → forget everything

TasteOS flips that.

Instead of asking you what you want, it quietly learns from:

Over time, it builds a personal taste fingerprint and uses that to suggest meals.

No “Italian healthy comfort food under 30 minutes” prompts.
No pretending taste is universal.

Does learning from behavior instead of prompts feel like the right abstraction?
Would you trust an AI to pick dinner after it’s watched you for a bit?
Where does this cross from “helpful” → “creepy” for you?
Does this solve a real problem, or just create a clever demo?

Not looking for growth hacks or launch advice here — just honest reactions from people who build products and think about UX, trust, and AI memory.

If you try it and hate it, that’s still useful.

Launching Wednesday 🚀
App is live now if you want to poke holes in it.

— Brady

6 views