Personal Knowledge System

A layered local knowledge system that separates immutable source records, reusable knowledge and AI-readable operating guidance.

The question

How can AI continue long-running work without exposing a private knowledge base or losing the path back to original evidence?

Approach

  1. Preserve incoming material as immutable source records.
  2. Write reusable guidance separately from raw evidence.
  3. Require every synthesized answer to retain a source path.
  4. Keep private storage separate from public website content.

Result

Raw records, synthesized knowledge and source checks now operate as distinct layers.

Evidence

69 maintained knowledge pages with traceable source paths.

Chat memory is not a knowledge system

AI conversations are fast, but sources, corrections and decisions often disappear with the session. This project keeps immutable raw records, reusable knowledge and operating rules as separate layers, with every synthesis pointing back to its source.

Keep source, synthesis and control separate

Raw records cannot be replaced by summaries; AI proposes rather than silently rewrites; routine queries are read-only; and audits check source drift, not only file formatting.

A record becomes reusable without replacing its source

New material is preserved unchanged before any synthesis is proposed. Only human-approved updates enter the knowledge layer; retrieval and source-drift checks remain separate operations.

  1. Preserve the raw record
  2. Propose a knowledge update
  3. Review it manually
  4. Query and audit without rewriting the source

Sixty-nine maintained pages remain traceable

Sixty-nine maintained knowledge pages can be reused across tools while retaining a route back to dated source records.

Evidence behind the case

The related Chinese article explains why the knowledge layer must remain independent from any one AI tool. Its English edition is still pending review.