CORE CONCEPTS

Knowledge with structure and provenance.

The ontology defines shared language. Knowledge describes the domain. Evidence connects to sources. Time makes changes explainable.

Domain model

Concepts, properties, relations, constraints

↓

Knowledge graph

Typed objects and meaningful connections

↓

Sources & evidence

Immutable source text, citations and provenance

A stable space for evolving knowledge

A space can begin empty. Once an Agent creates its model, knowledge follows that model. An upgrade creates a new generation, validates and migrates existing knowledge, then switches the current binding. Prior generations remain available.

Accepted knowledge keeps its provenance

Applied means the change passed authorization, structural, evidence-reference and concurrency checks and was admitted by the space policy. Agent-proposed content retains its origin for verification. Derived relations also retain their rules and premises.

Every update retains its history

An object can have different states across immutable revisions. Source digests identify snapshots that cannot be overwritten. Change receipts and run records explain when and why it changed.

See these concepts in the recorded ontology case

How does ontology-backed knowledge work with RAG?

RAG is useful for retrieving relevant passages from source text. An ontology runtime keeps stable concepts, entity relationships, constraints and history. An Agent can retrieve source text first, then persist reusable understanding as structured, evidenced knowledge.