I built a Read.ai connector for Cognee. Importing once wasn’t enough

Description and use

A Read.ai connector for Cognee that keeps meeting reports up‑to‑date by fully reconciling upstream data and incrementally updating downstream storage.

Problem

Meeting reports imported into Cognee become stale when later edited, because the connector only imports once and does not detect changes.

Outcome

The connector successfully keeps the Cognee database in sync with Read.ai reports, handling updates, deletions, and permission errors while avoiding unnecessary re‑processing of unchanged documents.

Workflow

• Select meeting IDs or a date range and optional summary/transcript/metrics inclusion
• Run sync_read_ai() which ingests, recovers, processes with Cognee, and verifies completion
• During sync, re‑fetch selected reports, render content, compute hashes, compare to stored hashes
• If hash differs, re‑process and replace document; if report missing, delete document
• After successful ingestion and verification, advance the successful checkpoint

Evidence

importing meeting reports is useful, but those reports do not stop changing once they reach your database.

Source

full upstream reconciliation with incremental downstream processing

Source

The successful checkpoint comes last—after ingestion, Cognee’s processing step, and verification.

Source

149 passing tests on Python 3.12.15, with no failures, errors, or skips.

Source