Connecting Intercom chats and Zoom call recordings into an automated answer library: which webhooks are used?

To ingest resolved sales and support interactions into an automated answer library, configure the `conversation.admin.closed` webhook in Intercom and the `recording.transcript_completed` webhook in Zoom to send JSON (JavaScript Object Notation) payloads upon event completion.

Customer interactions across support chats and video calls contain high-intent buying signals that disappear once a session concludes. Commercial teams spend between $100 and $300 an hour answering identical inquiries across private channels, which leaves high-value institutional knowledge trapped in ticket silos instead of driving organic search discovery.

If you only do one thing: Target closed interactions rather than live streams by filtering specifically for resolved chat threads and finalized call transcripts to prevent ingesting partial, unverified, or incomplete answers.

1. Configure Intercom event triggers: Register an endpoint for the `conversation.admin.closed` webhook inside your Intercom Developer Hub to export complete conversation threads with customer queries and validated agent replies as soon as a ticket closes.
2. Configure Zoom transcript webhooks: Enable the `recording.transcript_completed` event within a Zoom Server-to-Server OAuth (Open Authorization) application to automatically receive full VTT (Web Video Text Tracks) speech transcripts when cloud processing finishes.
3. Filter intent and parse question-answer pairs: Route incoming JSON payloads into an ingestion pipeline that classifies queries across 3 core commercial themes—pricing, technical specifications, and competitor comparisons—while stripping personal identifiable information.
4. Stage drafts for verification: Funnel structured question-and-answer pairs into a centralized dashboard with a two-layer review workflow so internal subject matter experts inspect and approve every draft before it goes live.
5. Publish structured assets to owned domains: Push approved answers directly to a custom domain or a dedicated `/answers` subfolder within a 7-day onboarding timeline to establish permanent, sourced assets for search discovery.

• Watch out for: Ingesting real-time creation triggers like `conversation.user.created`, which captures raw, incomplete customer questions before an expert has validated the solution.
• Watch out for: Publishing raw Zoom transcripts without automated redaction, which risks exposing sensitive customer data or PCI-DSS (Payment Card Industry Data Security Standard) payment records.
• Watch out for: Skipping the human verification step, which risks indexing outdated pricing tiers or unverified operational commitments on your public domain.

Set up a secure HTTPS (Hypertext Transfer Protocol Secure) endpoint in your middleware to listen for `conversation.admin.closed` and `recording.transcript_completed`, then test the workflow with 5 closed customer interactions to calibrate question extraction filters.