Auto-retrain
Auto-retrain watches source URLs and re-indexes them when content changes — so assistants answer from up-to-date pages without manual re-import.
Plan availability
Section titled “Plan availability”| Plan | Auto-retrain |
|---|---|
| Free, Starter | Not available — re-import or re-crawl manually |
| Standard, Pro, Enterprise | Available |
See Plans compared for full feature matrix.
What auto-retrain does
Section titled “What auto-retrain does”For URL imports and website crawl documents, Chatevo periodically checks the live page. When content differs from the stored snapshot, it:
- Fetches the updated page
- Creates a new snapshot
- Re-runs extract → chunk → embed → Qdrant
- Replaces active chunks for that document
Uploaded files (PDF, DOCX, TXT, HTML) are not auto-polled — upload a new version or replace the file to refresh.
Step 1 — Enable auto-retrain
Section titled “Step 1 — Enable auto-retrain”- Open a knowledge base with URL or crawl content.
- Go to Auto-retrain (or Settings → Auto-retrain).
- Toggle Enable auto-retrain on.
- Set check interval if exposed (e.g. daily, weekly) and save.
Step 2 — Monitor re-indexing
Section titled “Step 2 — Monitor re-indexing”Re-trained documents show Processing then Indexed in the document list. Failures (login wall, 404, empty page) appear as Failed — fix the source URL or disable auto-retrain for that document.
Enable boost_recent in Tune search settings so the latest snapshot ranks higher when multiple versions exist during transition.
Best practices
Section titled “Best practices”- Use auto-retrain on help centers and policy pages that change often
- Keep crawl selections focused — fewer pages means faster checks and lower cost
- After a major site redesign, run a fresh website crawl to discover new URLs
- Test in Test chat after large content updates