The Euro Banking Association (EBA) today announced the publication of a new report by its Open Finance Working Group (OFWG) titled “Supercharging embedded finance with agentic AI”. The aim of the report is to identify use cases in embedded finance where the leveraging of agentic AI would be of considerable benefit for financial institutions.

The report shows that agentic AI is poised to impact embedded finance in three distinct areas: enhancing customer experience and personalisation, automating complex operations to reduce cost and dynamically managing risk through adaptive, real‑time decision making. It follows that agentic AI will sit in specific spheres in the embedded finance stack: agentic interfaces that interact with customers in natural language, agentic orchestration that coordinates workflows and service calls across multiple parties, and agentic domain expertise that encapsulates the product logic, pricing rules, risk models and compliance policies of regulated institutions. 

“In preparation of this report, the Open Finance Working Group identified over 40 use cases across the embedded finance customer journey and developed an ‘agentic AI fit framework’ to assess them. The framework balances process and decision complexity, risk management implications and feasibility. We then shortlisted five use cases with high potential for adding value,” says Edwin Sanders, Chair of the EBA’s Open Finance Working Group (OFWG). “However, it is important to remember that organisational readiness will strongly influence the pace and success of agentic AI adoption.”

The five use cases discussed in the report include an agentic cashflow optimiser, procurement/commerce optimiser, know your customer/business (KYC/KYB) onboarding guide, payment reconciliation assistant, and product configuration manager. Five success factors ensure safe deployment: real‑time API access, end‑to‑end auditability, clear accountability and liability allocation, “human in command” checkpoints for critical decisions, and explicitly bounded decision rights for agents. However, the report highlights that introducing agentic AI in embedded finance carries new risks and requires substantial modernisation efforts. Following a clear “human in command” principle that ensures staff retain meaningful oversight and decision rights for critical actions will be key to building trust among employees, customers and regulators. 

At market level, three collaborative priorities matter: interoperable data and API standards for agentic embedded finance, cross‑industry digital identity frameworks for entities and agentic systems, as well as sector‑wide ethical and safety principles aligned with the EU AI Act and international guidelines. By progressing along the proposed agentic AI roadmap, industry players can ensure that agentic AI can be implemented safely – and supercharge embedded finance at scale.