Read with the operating model in mind.
Each section maps an idea back to the same private-AI pattern: boundary, source trail, review, and deployment.
Start with the data boundary
List the repositories, folders, case files, claim files, client documents, and policy sources the system may touch. Then list what must remain excluded.
The best early design decision is deciding where AI is allowed to operate and where it is not allowed to reach.
Map users, roles, and reviewers
Private AI is only as strong as its access model. Match retrieval and agent permissions to existing teams, matters, clients, departments, or programs.
For consequential work, name the people who must approve drafts, exports, summaries, or tool actions.
Define evidence requirements
Regulated teams need answers that can be inspected. Require source citations, prompt history, tool-call logs, and approval records from the beginning.
That evidence layer is what turns AI from a black box into an operational system your team can review.