← Continues from Why

How

How living knowledge is actually built.

Why made the case for a single truth two readers can act on. How is the architecture that makes that true — nine concepts, one structure.

Two ideas everything else stands on.

Living knowledge needs a unit it can be made of. The unit is the corpus.

Everything below is built out of corpora. Nothing below works without them.

Knowledge happens somewhere.

A corpus with no context is a file with extra steps. Projects and sessions give it one.

A project remembers. A session contributes. Neither resets when the call ends.

Knowledge has to remember how it got here.

A fact with no date is a guess wearing a costume. Graph Memory links it. Temporal Knowledge dates it.

Most systems store the current state. This one stores how it became current.

Knowledge has to keep working after launch day.

A corpus that stops watching the world goes quietly wrong. Discovery / Publishing keeps it current. Evidence keeps it honest.

Passive RAG

  • Answers questions
  • Depends on existing sources
  • Becomes stale
  • Rarely explains what is missing

Discovery / Publishing

  • Detects missing coverage
  • Discovers candidate sources
  • Refreshes memory selectively
  • Keeps humans in control

A system that can’t tell you what it doesn’t know yet isn’t trustworthy. It’s just confident.

None of this works alone.

A corpus, a project, a session, a graph, a published artefact — each is real on its own. None of them is the architecture. The constellation is.

The architecture, assembled

PROJECTCorporaProjectsKnowledge SessionsDiscovery / PublishingGraph Memory

This is what Why asked for. This is how it’s built.

See it assembled, not just explained. The product page shows every one of these concepts at work, end to end — and the API shows them callable, not just visible.