The Future Bank Council convenes a Tactical Skill Workshop focusing on Building effective and responsible AI agents for financial services.

As financial institutions explore the use of agentic AI, attention is shifting from experimentation towards how AI agents can be designed, deployed and governed in practical business environments. Understanding what distinguishes an agent from a conventional workflow, and when simpler approaches may be more appropriate, is becoming increasingly important.

Institutions must determine how to translate business requirements into effective agent designs. This includes defining clear outcomes, selecting appropriate tools and information sources, establishing rules and controls, managing memory and context, and designing systems that can be tested, measured and improved over time.

The session examines how financial institutions can move from basic AI agent concepts to practical implementation while controlling cost, complexity and risk. It will also consider how organisations can validate agent outputs, manage security and governance requirements, and determine when a single agent architecture is sufficient versus when multi agent approaches may be appropriate.

In this interactive session, we will examine:

· How financial institutions can identify where AI agents add value, and when workflows or simpler AI patterns may be more appropriate

· What design, tooling, testing and measurement practices are needed to build reliable AI agents and demonstrate measurable outcomes

· How banks can manage data, security, governance, cost and human oversight as AI agents move from experimentation towards production