Turning Meeting Transcripts into an AI Knowledge Graph with tl;dv & Cognee

Description and use

The tl;dv and Cognee connector consolidates meeting recordings into a single Markdown document, extracts entities, and builds a knowledge graph, enabling precise entity‑grounded queries.

Problem

Engineering and product teams lose vital context from meetings because recordings and notes are scattered and not easily searchable.

Outcome

The connector transforms raw meeting data into a knowledge graph, enabling entity‑grounded queries such as identifying who volunteered for a project and their blockers, which plain vector search cannot resolve.

Workflow

• 1. Use tl;dv API to fetch meetings, transcripts, and notes via /meetings, /meetings/{id}/transcript, /meetings/{id}/notes.
• 2. Synthesize data into a single Markdown document per meeting.
• 3. Annotate the document with DOCUMENT_SOURCE_ATTR="tldv_document" for Cognee ingestion.
• 4. Add the source to Cognee dataset and run cognify to extract entities and build the graph.
• 5. Query the knowledge graph with natural language questions via Cognee search.

Evidence

Engineering and product teams spend dozens of hours every week in meetings. Crucial technical decisions get debated, architectural tradeoffs get resolved, and action items get assigned.

Source

If you store meeting transcripts as plain vector chunks, multi‑speaker reasoning falls apart.

Source

Our connector fetches and synthesizes these into a single unified Markdown document per meeting

Source

By annotating the dlt resource with DOCUMENT_SOURCE_ATTR = "tldv_document", Cognee treats each meeting as a document node, running its NLP extraction pipeline over the executive takeaways and dialogue.

Source

uv run pytest packages/connector/tldv/tests/test_tldv.py -v

Source