meeting-ingestion

Meeting transcript ingestion works by parsing the transcript to extract attendees, dates, and actions, then creating a dedicated brain page with bidirectional back-links to related people and company pages. It operates sequentially through phases: parse, create, enrich, propagate, and timeline-merge.

You can create a meeting page by providing a meeting transcript to the agent and triggering the meeting-ingestion skill. The skill then automatically parses the transcript, creates the dedicated brain page, enriches attendees, and begins entity propagation across your knowledge graph.

Yes, meeting ingestion enforces mandatory attendee enrichment and bidirectional back-linking to ensure complete traceability across your knowledge graph. It also merges timeline data across all mentioned attendees, companies, and concepts to maintain a coherent activity history.

Meeting ingestion handles multiple attendees by enriching each person mentioned in the transcript and creating bidirectional back-links between the meeting page and their individual pages. It propagates entities and merges timelines across all pages to maintain a connected knowledge graph.

Meeting ingestion operates in deterministic phases: parse to extract data, create to generate the meeting page, enrich to enhance attendee data, propagate to spread entities, and timeline-merge to integrate activity history. This structured approach ensures complete traceability across the knowledge graph.