This guide covers Intercom as a data source for training. For connect, conversations, Assist Mode, channels, and handoff, see the Intercom integration.
Set up the sync
Open Data Sources
Go to AI Training → Data Sources in the OpenCX dashboard.
Add an Intercom source
Click Add source → Intercom. Enter:
| Field | Value |
|---|---|
| Access token | An Intercom personal access token with read permissions. Can be the same token used for the ticketing integration or a separate one scoped to read. |
| Start date | The earliest date to sync data from, in YYYY-MM-DDTHH:mm:ssZ format. |
You can reuse the access token from the ticketing integration. It needs permission to read conversations and contacts.
Start the initial sync
Click Start sync. The first run pulls data from your start date forward. Depending on volume this takes minutes to a few hours. Progress is visible in the source row.
Verify data is indexed
Once sync completes, open AI Training and search for a known topic. Content should appear with
Intercom as the source.What gets synced
Conversation segments
Conversation segments
Intercom customer segments are synced, giving the AI context about how your contacts are grouped and what patterns exist across your support history.
Incremental updates
Incremental updates
After the initial sync, subsequent runs only fetch data that changed since the last run. New conversations and updated segments are picked up automatically on the next scheduled run.
Re-sync and disconnect
- Re-sync: click Sync now on the Intercom source row in Data Sources to trigger an immediate incremental sync.
- Disconnect: click Remove on the source row. This deletes the synced data from OpenCX’s training index. The original data in Intercom is not affected.
Related Documentation
Connect Intercom
Ticketing integration setup — separate from Knowledge Sync.
Conversations
How the AI uses conversation context at runtime.
Intercom overview
What else the Intercom integration covers.
Troubleshooting
Sync not completing, data not appearing in answers.