Human-Agent Knowledge Roles

Humans should steer evidence and meaning while agents handle repeatable filing, integration, linking, and consistency maintenance.

Divide durable knowledge work according to comparative strengths. Humans select sources, direct inquiry, judge significance, and resolve consequential interpretations. Agents perform repeatable bookkeeping: distillation, filing, cross-referencing, propagation of revisions, and consistency checks.

The split realizes the Memex idea of a private, curated store linked by associative trails: humans choose the trails and the focus, and the agent supplies the tireless maintainer the original concept lacked.

Encode this division in the knowledge maintenance operating schema, including which mutations require human review. The boundary keeps automation focused on the maintenance cost curve without treating fluent synthesis as authority over source quality or meaning.

---

Sources

• LLM Wiki, Why this works and Note