AI SOP knowledge system
Planning an AI system for a healthcare-related foundation, so that internal procedures and working documents are easy to find, stay up to date and do not leak sensitive data.
This project started with a talk. I was invited to speak about how RAG can be used in day-to-day work, and about the privacy and data governance questions to watch for when adopting AI. About 80 people attended, online and in person. After the talk, the two sides went on to discuss requirements and planning for an AI SOP system.
With everyone in the room after the talk.
The problems they face
| Pain point | Planned direction | Expected improvement |
|---|---|---|
| Procedures and internal knowledge are scattered | AI question answering with sources attached | Finding the right process and rule faster |
| Information goes out of sync after documents are updated | Version management and a knowledge maintenance process | System content keeps up with the latest procedures |
| Personal and sensitive data are involved | Data handling, access control and boundaries for LLM use | Lower data risk when adopting AI |
My role
Taking part in requirements interviews, assessing technical feasibility and planning the system architecture. The main work is turning the business side’s still-vague needs into a technical plan that can be carried out.
Where it stands
Still in requirements interviews and planning. We keep confirming use cases, data scope and feature needs with the users, and will then assess a proof of concept and the direction for development.