Knowledge ingestion playbook

1. Prepare your sources

• Internal docs: Export from Notion, Confluence, or Google Docs as markdown or HTML.
• Tickets: Pull resolved tickets covering common scenarios.
• Public site: List the base URLs that should be crawled.

2. Upload documents

1. Go to Knowledge → Sources.
2. Drag and drop files or connect cloud drives.
3. Tag documents by brand, product, or policy area.
4. Set review reminders so owners revisit content on a schedule.

3. Configure the crawler

1. Add URLs for policies or help centers.
2. Limit depth and rate to avoid overloading your site.
3. Exclude pages with dynamic or personalized content.

4. Sync historical tickets

• Use the CLI or API to bulk import resolved conversations.
• Include resolution notes and macros to train AI responses.
• Optionally redact PII during import via built-in masks.

5. Validate results

• Use Test query to see how CORE AI cites sources.
• Flag inaccurate snippets for retraining via the feedback tool.
• Monitor freshness indicators to catch outdated content.

6. Automate updates

• Enable scheduled crawl jobs.
• Connect the Mesh recipe reindex-on-merge to trigger after knowledge PRs.
• Configure alerts when high-impact docs go stale.

Keeping your knowledge base fresh ensures AI suggestions stay on-brand and accurate.