"Onchain intelligence" gets used loosely. Here's what it actually requires once you try to ship it — notes from incubating ORBIT.
It's a data problem first
Before it is a compliance product, onchain regulatory intelligence is a data problem: ingest heterogeneous signals, resolve entities, and enrich transfers with the context that makes them meaningful. Get the data model wrong and no amount of policy logic saves you.
Three layers that matter
- Attribution — connecting addresses to real-world entities with graded confidence.
- Sanctions — screening against lists that change, with an audit trail of why something matched.
- Banking context — mapping exposure and flows into terms an institution's risk team already uses.
Keep the reasoning attached
The failure mode of regulatory tech is the black box. Every alert must carry its rationale so a human can defend the decision — to a boss, a board, or a regulator. Surveillance without a decision trace is just noise with authority.
The takeaway
Build the pipeline before the policy. The institutions that adopt onchain intelligence are the ones who can explain it. That principle shaped ORBIT's architecture more than any regulation did.