Hi Product Hunt 👋 Solo dev + grad student building memory for AI coding agents

Hey everyone

I'm Ripon — a grad-student researcher at the University of Utah and the solo developer behind projectmem.

What pulled me into this: my AI coding agent kept forgetting everything between sessions. Every morning it re-read the same files and cheerfully re-suggested a fix I'd already tried last week — one that had already failed. The model wasn't the problem. The missing memory was.

So I built projectmem: a local-first memory + judgment layer that lets an AI agent remember what was tried, what failed, and what was decided — and warn you before you repeat a known dead-end. 100% local, MIT, no telemetry, no cloud. It's on PyPI and the MCP registry now, and there's even a research paper behind it.

Honestly I'm here to learn from other builders as much as to share. Solo building is humbling — shipping the code is 20% of the work; the docs, the launch, the community is the other 80%.

If you're working on AI tooling, developer experience, or just shipping solo — I'd love to connect 🙏

Question for the room: what's the one thing you wish your AI assistant actually remembered?