Cleafy, the fraud detection and response platform for banks, has launched Nyx – the first autonomous fraud investigation platform built specifically for financial institutions. Nyx deploys AI agents that investigate every fraud signal automatically, reconstruct full attack campaigns, deliver complete forensic investigations, and generate optimised detection and response rules – all in less than four minutes.

For most banks, the problem is no longer detecting fraud. It is investigating it fast enough. The investigation stage has become the new point of risk – where fraud campaigns expand, losses accumulate and operational teams fall behind volume.

Nyx introduces a new operating model for fraud investigation, delivering immediate advantages for financial institutions:

• Faster decisions on fraud signals. Nyx dramatically reduces the time needed to analyse and close fraud investigations to less than four minutes, down from more than four hours when done manually.
• Every fraud signal investigated automatically. Most fraud teams investigate only a fraction of the signals generated by their systems. The remainder are closed quickly or triaged away due to capacity limits. Nyx allows financial institutions to investigate 100% of fraud cases with the same level of forensic depth, something manual teams cannot scale.
• Earlier identification of coordinated attacks. Nyx has already identified coordinated attacks up to fifteen days before the first fraudulent transaction occurred.
• Automatic defence optimisation. For every attack pattern it reconstructs, Nyx generates optimised detection and response rules and rationalises the institution’s entire existing ruleset – eliminating redundancy, resolving conflicts, and strengthening coverage without manual rule maintenance.

Human analysts supervise the system rather than handling each case individually, and their feedback – together with new evidence from live cases – provides continuous learning for Nyx’s AI agents. Nyx generates forensic investigations in an analyst’s working language, which allows investigators to review findings immediately without translation.

Matteo Bogana, chief executive officer at Cleafy, said: “Fraud has become an industrialised operation. Increasingly, AI is being used to orchestrate and accelerate fraud attack campaigns at a scale and speed that manual defences cannot match. Yet investigation workflows have remained largely manual, and fraud systems generate more signals than investigation teams can realistically analyse. Nyx changes this, leveraging AI to fight AI. We’re creating a fraud defence that matches the threat.

“Nyx is the beginning of autonomous fraud operations, from investigating and managing attacks in minutes to dismantling them before the first fraudulent transaction occurs.”

Designed for regulated environments, Nyx has full logging and traceability that aligns with regulatory frameworks such as DORA and NIS2.

Nyx is already operating at three European banks collectively serving over 30 million digital banking customers:

“The quality of the investigation report was beyond what we expected.” Fraud operations lead, European bank.

“We spend 4–5 hours per case. Nyx completed the same investigation in under 4 minutes.” Head of fraud, European financial institution.

“When we saw the report in our own language, everything changed.” Chief information security officer, Southern European bank.