Case study
Building a context foundation for AI
A shared source of truth for engineering and product, written for people and AI tools alike, that saved tens of hours.
Context
AI tools were part of everyday engineering work, but they knew nothing about our product, our systems or our language.
Challenge
Without context, AI gave generic answers. People also kept spending time on the same questions, on debugging, and on working out what a ticket actually meant.
Approach
- Developed a context foundation template: a simple, shared structure for describing a part of the product.
- Used it to document one team’s module, so it became a source of truth for both engineering and product.
- With the AI champions, started rolling the template out to all seven engineering teams.
Outcome
It saved tens of hours on questions, debugging and understanding tickets. It also made it easier to explain the product to both technical and non-technical colleagues.
What I learned
Context written for AI turned out to be just as useful for people. Good context is simply good documentation with a clear reader in mind.