research

AI Cooking Coach (Process Debrief)

Research into a cooking coach that reads probe logs and step times, then tells you why a batch went wrong.

This started as “what if a camera and a probe thermometer could coach you through a recipe?” and turned into something narrower. The interesting part isn’t watching you cook. It’s the debrief afterwards: looking across many batches and finding the cause that keeps coming back.

The write-up on how I got here is in What If the Debrief Is the Product?.

Where It Stands

  • Market research: done. 12 segments researched, each one checked by a second pass told to refute it. 47 ideas went in, and the survivors were scored and ranked.
  • Cooling Debrief demo: done, on generated data. It scores commercial cooling logs against the Food Code cooling rules and drills into why batches failed.
  • Video step timing: in progress. Early tests say it only works for long steps.

Demos

The two demo pages are getting a cleanup pass and will go up here when they’re ready.

What’s Next

Testing whether a real person in food service can find the planted problems in the Cooling Debrief in ten minutes, with no hints. If they can’t, the page isn’t doing its job yet.

← All Projects