voice-post-call

Voice call transcripts are ingested to automatically generate a structured brain page with cross-linked entities and a concise summary for quick decision-making. The workflow processes live or end-of-call transcripts and archives the original audio for later review.

The best way to archive voice call audio and transcripts is using a post-call pipeline that writes canonical records and links audio files to structured brain pages. This ensures idempotent repeated invocations and preserves privacy by not exposing raw transcripts beyond the brain page.

You can apply this workflow to both live voice calls and end-of-call ingestion. The pipeline processes transcripts in real-time or post-call, creating brain pages and summaries with entity cross-linking regardless of the ingestion method.

Voice post-call processing preserves user privacy by not exposing raw transcripts beyond the generated brain page. The workflow stores canonical records with audio references under meetings/ or voice-calls/ directories, maintaining traceability without compromising sensitive transcript data.

The voice call ingestion workflow is idempotent on repeated invocations, meaning repeated processing of the same call will not create duplicate records. It writes canonical records following established conventions, ensuring consistent archival and review across multiple pipeline triggers.