Glossary & Shared Vocabulary
Investigations involve the same names, organisations, and terms coming up again and again — often written differently each time. The glossary is Classifyre’s shared vocabulary, so that “Acme Corp,” “ACME Corporation,” and “Acme” are recognised as the same thing, by people and by AI agents alike.
What it’s for
Without a shared glossary, an investigation drifts: one person writes a full name, another writes an abbreviation, a detector reports a raw value, and nothing connects them automatically. The glossary fixes that by giving each real-world thing — a person, an organisation, a location, a reference number, or any other recurring term — one canonical entry with a list of aliases that all point back to it. Looking something up matches on the exact term, any of its aliases, or a close semantic match, so a search doesn’t depend on getting the wording exactly right.
Every entry also carries an entity type — person, organisation, location, reference, or general term — which helps keep searches and suggestions relevant.
Operator terms vs. agent terms
Anyone can add to the glossary, but not all entries start out equal:
- Operator-curated terms are entries you or your team define directly. They’re trusted immediately.
- Agent-proposed terms are suggestions from an AI agent that noticed a recurring name or reference worth tracking. These start unverified — clearly marked as agent-origin — until a person reviews and confirms them.
Critically, an agent proposal can never silently overwrite or replace an operator’s entry. At most, an agent can suggest an additional alias for a term you’ve already defined — it cannot redefine what you’ve established.
Why this matters day to day
A well-maintained glossary makes everything upstream sharper: fingerprint connections and inquiry matches read variant spellings correctly, case notes reference the same canonical entities, and Autopilot agents reason about “the same person” instead of treating each spelling as a new unknown. It’s a small amount of curation that compounds across every investigation you run.
Where this leads
The glossary is consumed everywhere findings, connections, and cases are worked — and it’s one of the things Autopilot agents can propose additions to, always subject to your review. See Autopilot & AI Assistance for how agent-proposed content is governed more broadly.