Brainbase

Most working knowledge is scattered across chat histories, email, document folders, note apps, task lists and AI conversations. Each tool holds a fragment, and none of them know about the others.

Search finds text, not meaning. It rarely answers the questions that matter later: what did we decide, who was involved, and where did that decision come from?

AI assistants make this more pressing. They are only as useful as the context they are given, and most of them start every conversation with none.

Brainbase separates what was captured from what was understood. Raw sources are archived unchanged and never edited. Everything else — normalised records, extracted entities, relationships — is derived from those sources and can be rebuilt from them.

Sources are ingested, normalised, deduplicated and classified. Brainbase then extracts the things people actually think in — people, projects, decisions, tasks, topics and documents — and records how they relate to one another.

The result is a knowledge store with provenance on every item and backlinks in every direction. It can be searched directly, written out as plain Markdown for Obsidian, or made available to AI agents as structured, persistent context.