Build a human-readable knowledge wiki from selected Microsoft Graph sources, with review controls visible in the content. We’ll trace a prototype that creates Markdown pages with schema-validated metadata for source identifiers, versions, owners, approval status, and review dates. The pipeline detects source changes, produces proposed revisions, and runs scheduled reviews that expose pending and overdue work. Technical examples address conflicting evidence, malformed output, and access revocation. We’ll explain how this approach complements Work IQ while retaining links to authoritative originals. Generated pages require their own access controls rather than assuming source permissions transfer automatically. Participants get a reference architecture, sample metadata, and tests for knowledge people can inspect and approve before agents reuse it.

Learning objectives
Design a selective Graph-to-Markdown pipeline with validated provenance, ownership, and approval metadata.
Implement change detection, scheduled reviews, and visibility into pending or overdue work.
Define access controls for derived content, including responses to source-access changes.