Our intelligence meta-ontologies are precise, machine-readable maps of the objects and categories in a system, consisting of terms, concepts, actors, actions, the relationships between them, and constraints on how they combine. We built four from the ground up, using intelligence management doctrine, data integrity standards, and desk and field research. Two of these are meta-ontologies - ontologies of ontologies - covering standards of evidence. Two are domain-specific ontologies that we developed for granular monitoring and tracing of geopolitical developments.

Graphs are the runtime knowledge base an AI system queries while it works. Our graphs are tightly focused collections of indexed terms drawn from loaded ontologies and catalogued sources, combining a fast local index for search with a standards-based store for portable exchange. Every source carries a cryptographic fingerprint together with the URL, timestamp, response status, and byte count it was captured with, so any term can be traced back to the record it came from. Two graph explorers ship with these: a local server for live inspection, and a single-file offline viewer that runs without the need for a network connection. Both carry chain navigation, comparison views, and freshness reporting, and export functions.