Corpus Plus Dialogue

Pattern · knowledge_systems · personal_truth · visibility: publishable

The working PKM is not a tagging system or a hierarchy. It’s corpus plus dialogue. Frictionless capture (journal, workshop mode) accumulates raw material without demanding atomization at point of capture. Dialogue — pointing Claude at the corpus and talking through it — does the atomization after the fact, in the mode the brain actually works in (conversation). Structure emerges from the back end, not imposed at the front. This is the positive architecture of what actually works: the journal bypasses the atomization demand entirely; the dialogue satisfies it later. Distinct from P.capture-tools-fail (which names the failure) and P.system-must-fit-not-teach (which names the fit principle): this is the positive architecture. Instances: Diarly + Claude extraction process, linguistic analysis chat, writing voice chat, knowledge graph project itself.

Connections

source: Linguistic analysis of writing files (2026-02-13) · extraction: hand (schema v3.1)