โ† AI library

Editorial recipe โ€” not benchmarked โ€” reviewed September 1, 2026

Summarize sensitive documents without a cloud-model handoff

For teams authorized to process the documents locally and able to verify every material claim.

Stack
Docling (local) โ†’ Ollama + qwen3.5:9b โ†’ Obsidian โ†’ Human reviewer

Run this recipe

Work through it here

Progress stays in this browser. The downloaded Markdown kit works in any notes app or repository.

0/5 complete

Procedure

  1. Copy only authorized files into an isolated, non-synced working directory; record file hashes, the purpose, and a deletion date before processing.
  2. Run Docling locally to export Markdown and structured JSON, then compare tables, footnotes, page order, and OCR-sensitive passages against the original documents.
  3. Run qwen3.5:9b through the local Ollama endpoint; save the model manifest digest and prompt, and request a claim table tied to Docling source locations rather than a free-floating summary.
  4. Have a human open every source location behind a material number, date, obligation, or recommendation and mark extraction errors or unsupported inferences.
  5. Store the approved brief, claim table, model digest, and exceptions in Obsidian; delete working copies on the recorded date.

Acceptance artifact

A locally produced brief with file hashes, model digest, source-linked claim table, extraction exceptions, and human verification status

Do not use it blindly

Local execution is not automatically private: disable cloud endpoints and sync, verify the process network policy, and never use a small local model as the final reviewer for legal, medical, financial, or safety-critical decisions.

Evidence and setup