Raw customer data stays inside each institution. Fraud intelligence moves across the network. Alerra never holds a bank’s customer database — it connects what institutions have each already concluded.
A customer attempts a $48,000 transfer to a beneficiary created days earlier. Bank A’s own fraud engine scores it, and the score is not high enough to stop a payment on its own.
Bank A’s Alerra Gateway derives a narrow, privacy-preserving object and publishes that. The customer, the account and the transaction history never leave the bank — only this does.
Illustrative object. Entity tokens are pseudonymous identifiers, not customer or account identifiers.
None of these institutions knew about the others. Each held one fragment, none of them conclusive on its own.
Nothing about the transaction changed. What changed is how much context existed to judge it against.
Illustrative risk assessment from synthetic data.
Bank A confirms the scam. That confirmation is the most valuable signal in the system, because an outcome is a fact rather than an inference — and every connected institution has it immediately.
The same cycle, on synthetic data, continuously. Every node is an entity some demo institution has observed; every link is an observed relationship.
A score a bank cannot interrogate is a score a bank cannot act on. Pick any synthetic entity and read exactly where its risk came from.
Data stays. Intelligence moves. Evidence compounds.