Case study · AI
A pipeline that turns rough meeting notes into requirements, decisions, risks and open questions, and remembers every earlier session.
- 5outputs from every meeting
- Allearlier sessions kept in memory
- 1summary posted to the team
The problem
Decisions made in meetings were getting lost, and each new meeting started without the context of the last one.
What we built
- 01
A deterministic pipeline that extracts requirements, decisions, risks and clarification questions.
- 02
Project memory with vector embeddings, so each run retrieves what was decided before.
- 03
Summaries posted straight to the team channel.
The result
Every meeting produces the same structured record, built on everything that came before it.
Next case study
A multi-tenant CRM, built for many companies at once