conversation-knowledge-flywheel

You convert AI chat transcripts into structured knowledge by applying nightly automation that inventories files, infers project groupings, and renders writer-ready packages. The process routes project references to a registry for direct visibility.

Conversation mining for knowledge management is the process of transforming raw chat transcripts into discoverable, project-anchored knowledge artifacts. It applies manifest generation and surveillance scans to surface priority candidates and daily learning suggestions.

Nightly automation for transcript processing requires qmd, node, npm, and python3 dependencies. You run the workflow to execute QMD healthchecks, embed candidates for high-recall retrieval, and handle degraded-run reports if repairs fail.

Yes, the conversation knowledge flywheel works with Obsidian by routing project references to a Brain-side 05-PROJECTS registry and performing commit and visibility verification so generated drafts become Obsidian-visible notes.

QMD-based retrieval for conversation transcripts can experience degraded runs if healthchecks or repair attempts fail. The workflow handles these limitations by generating explicit degraded-run reports to indicate when high-recall candidate retrieval is compromised.

Yes, you can generate research seed candidates from daily AI conversations for NotebookLM reinforcement. The workflow produces 1-3 knowledge drafts, a daily suggestions block, and optional research seed candidates from yesterday's transcripts.