The ECB's SupTech Toolkit — 14 AI Applications Now Used by Over 3,500 Bank Supervisors Across the Eurozone

At the end of 2019, European banking supervision under the European Central Bank (ECB) began building its own supervisory technology rather than relying solely on manual review of the reports banks submit — an effort organised around a dedicated SupTech Hub.

Over the following three years the Hub delivered 14 applications and platforms now used by more than 3,500 people across the ECB and national banking-supervisory authorities in the euro area. Named tools include a natural-language query tool that lets supervisors without coding skills pull specific data points from the ECB's internal "data lake"; Athena, which translates and analyses supervisory documents and external media coverage; GABI, which generates and optimises statistical/regression models at scale; NAVI, which draws network diagrams of ownership structures and interdependencies among supervised banks; Heimdall, which processes large volumes of information to support "fit and proper" assessments of bank managers; and Medusa, which checks the consistency of reports drafted after internal-model investigations.

By 2024 the ECB had identified more than 40 further potential uses for generative AI in banking supervision. The bank describes its approach as human-in-the-loop: supervisors read the context, assess the AI-generated content, and provide feedback that feeds an iterative improvement process, rather than letting the models make supervisory decisions outright.

The sources reviewed document adoption and usage at real scale — applications built, users served, use cases catalogued — but do not publish independent metrics on outcomes such as time saved, errors avoided, or risks caught earlier as a result of the tools, nor a public audit mechanism for how the tools are used.