=^ w ^=
← ALL PROJECTS

Nibbler

2026 · AI
PYTHON · LLM APIS · MCP · POSTGRESQL · VECTOR SEARCH
VIDEO NOT AVAILABLE :c

NOTES

Everyone I know has the same problem: a friend sends you a restaurant, you say "oh that looks good", and then it disappears into a group chat forever. Nibbler goes and finds those mentions across Instagram, Discord, WhatsApp and SMS, parses them with an LLM, and keeps them somewhere you will actually look.

The interesting half is the matching. It builds a preference profile out of your order history and spending patterns in Postgres with vector search, then uses MCP connectors and a calendar API to pull in real restaurant data and factor in when you are actually free. You get menu suggestions and a cost estimate per visit rather than a list of names.

There is also a sponsored-placement layer where restaurants can bid on recommendation visibility, which is the part that could pay for itself.