Skip to main content
Configure the source at AI Training → Data Sources.
Microsoft grants these permissions across your whole tenant. OpenCX reads only the site and folder you enter, but the app itself can read other sites, so keep its secret private. Files in the synced location can appear in customer answers as soon as they sync.

​
Set up the sync

1

Register an app in Microsoft Entra

In the Microsoft Entra admin center, go to App registrations → New registration. Note the Application (client) ID and Directory (tenant) ID from the app’s overview page.
2

Grant read permissions

Open API permissions → Add a permission → Microsoft Graph → Application permissions. Add Sites.Read.All and Files.Read.All, then click Grant admin consent. A Microsoft administrator has to do this step.
3

Create a client secret

Open Certificates & secrets → New client secret and copy the secret’s Value. Microsoft shows it only once. Note its expiry date: syncs stop when the secret expires.
4

Add the source in OpenCX

Go to AI Training → Data Sources and click SharePoint.
5

Start the sync

Click Connect. OpenCX checks the credentials, the site and the folder, then starts the first sync. Large libraries take longer.
6

Verify in AI Instructions

Open AI Training → AI Instructions. Your files appear under Imported from SharePoint, each linked to its file in SharePoint.

​
What gets synced

PDF, Word, PowerPoint, Excel and Markdown files in every document library on the site, including subfolders. When you set a folder, each library is searched for that folder and libraries without it are skipped.
New files arrive as public knowledge, so the AI can use them right away. To keep a file or folder away from the AI, use the eye icon on an item or Hide all from AI on the folder. Your choice is kept on later syncs.
Each sync picks up new and edited files. Files deleted in SharePoint, or moved out of the synced location, leave the index after the next sync.

​
Limits

SharePoint overview

What this source does and doesn’t.

Troubleshooting

Connection errors and missing files.

Website crawler

For content reachable at a public URL.

Connect a knowledge source

All sources.