Make a document library Copilot-ready: OCR, metadata and permissions check

Prepare SharePoint content for Microsoft 365 Copilot and Copilot Studio agents: OCR scans, tag metadata, and flag over-shared folders.

TriggerManual (one-off backfill) and When a file is created (ongoing)
ConnectorsSharePoint · Adobe PDF Services · AI Builder (optional) · Microsoft Teams
CategoryIT & admin

The problem

Copilot answers poorly because content is scanned images, untagged, or shared too widely.

The outcome

Searchable, tagged content with permissions checked — better Copilot answers and lower oversharing risk.

Before you start

  • A pilot library, ideally one department's policies or contracts.

Build it step by step

  1. Find image-only PDFs — Loop files and test for text; queue image-only PDFs for OCR.
  2. OCR — Run Adobe PDF Services OCR and replace the file (keep version history).
  3. Tag — Use AI Builder or a simple classifier prompt to set Document Type and Department metadata.
  4. Check sharing — Flag folders with 'Everyone' or org-wide links and send to the site owner for review.
  5. Report — Summarise files processed, tagged and sharing issues found.

Connector action names are shown as they appear in Power Automate at the time of writing; Microsoft and Adobe occasionally rename actions or add new versions.

Error handling

  • Run the backfill out of hours and in batches to avoid throttling.

Licensing

  • Adobe PDF Services per transaction; AI Builder credits (premium).

Variations

  • Use SharePoint Advanced Management for tenant-wide oversharing reports.

Common in: Government · Legal · GLCs