About
The open library of financial-services agents.
Cadre is a public reference for AI agents built to do real work inside banks, lenders, and payment firms. Every agent here is a working system, not a demo. You can read how it is put together and see what it did on real runs against representative cases.
What is in the library
The library groups agents by the line of business they serve (financial crime, compliance, lending, markets, fraud, disputes, and governance). For each agent you can inspect its anatomy: the model it runs on, the tools it can call, the connectors it reaches into, the skills it exposes to other agents, and the guardrails that constrain it. The catalog is the same registry that powers the console; nothing here is a mock-up.
What a replayed run shows
A replayed run is the recorded trace of an agent working a single case: the steps it took, the tools it called, what it read, and the decision it reached. The cases are representative domain fixtures; the runs are captured from real model executions (not scripted walkthroughs), then frozen so anyone can step through the same reasoning we saw. Replays are always labeled as replays; nothing on this site executes live. The receipts are the point. If an agent claims to triage an alert or adjudicate a dispute, the run shows its work.
Who builds it
Cadre is built by Aarvion. The agents deploy governed and run on your own cloud, so the model calls, the data, and the audit trail stay inside your boundary. This site is the open shop window; the deployed product is the console. There is no pricing page and no signup here. If you want to talk, the footer has an address.