DocsGPT connects to SharePoint and OneDrive through its connectors, indexes the documents inside your Cloud, Managed or On-premises deployment, and answers with citations to the source file and page. Microsoft 365 remains the system of record; DocsGPT reads from it and, on Managed and On-premises, keeps the index, embeddings and model inside your environment. Each person signs in with their own Microsoft account through Entra ID, and sync runs daily, weekly, monthly or on demand. Your admins decide in Admin › Connectors whether the connector is enabled for everyone. The connector needs reachability to your Microsoft 365 tenant, so an Air-gapped deployment ingests exported files by upload instead.
| | Read by the connector | Notes |
|---|
| OneDrive | Files and folders in the signed-in person's OneDrive | Browse and pick files or folders when adding the source |
|---|
| Shared with you | SharePoint and OneDrive files and folders shared with that person | Work or school accounts; same browse-and-pick selection |
|---|
| File formats | Office files (docx · doc · xlsx · xls · pptx · ppt), PDF, text (txt · md · rst · html · json · rtf · epub), CSV and images (png · jpg); 30+ formats in all | Scanned PDFs and images pass through OCR |
|---|
| Metadata | File name, path, modified date | Shown in the citation panel |
|---|
| Permissions | The connector reads only what the signed-in Microsoft account can open; inside DocsGPT, access follows SSO (OIDC), roles and per-source sharing | Sources are shared with teams and people inside DocsGPT |
|---|
SharePoint and OneDrive only; no Teams connector.