Organisations should adopt a strategic, phased approach to incorporating Agentic AI. This involves identifying business value for potential use cases, defining detailed personas and goals, defining risk appetites, updating risk assessment processes, and implementing controls to manage AI-specific risks effectively. Starting small and refining the approach will ensure scalability and trustworthy implementations.
Here are the practical steps to get started with AI Agents, according to the latest IBM paper called Agentic AI in Financial Services – Opportunities, Risks, and Responsible Implementation.
1. Determine your strategy & plan
a. Define your strategy in determining the use of AI agents and how this will work with your current people, tools and processes
b. Identify business value for key potential use cases and applications
c. Define detailed personas, goals, constraints, and expected outcomes for identified use cases and workflows at the most granular level
d. Implement the use case and its controls at the same time
e. Evaluate use case and controls performance prior to deployment, including FinOps KPIs
2. Ensure safety & trust
a. Define your AI ethics principles and determine the kind of relationship your company wants to have with AI
b. Define clear functional and non-functional requirements to operationalise your AI principles
c. Uplift your risk and controls library for Agentic AI
d. Establish a centralised, standardised repository of use cases, consumable to both technical and nontechnical stakeholders
e. Stand up a cross-organisational AI literacy program
f. Ensure comprehensive end-to-end risk management processes. Shift the risk assessment process left, strive to automate as much of it as possible
g. Incentivise a risk-oriented innovation culture, with easyto-use risk identification guidance, and an operating model with clear accountabilities
h. Implement real-time monitoring
i. For each use case, identify risks to govern the AI agents
and/or agentic systems, and design effective controls to
mitigate. Determine the level of rigour required to mitigate
for each component of the agentic system
3. Determine approach to scalability
a. Determine the strategy and approach to building AI agents leveraging agent frameworks to accelerate scale. Whilst agentic frameworks help accelerate development, they may also potentially introduce unnecessary complexity and abstractions
b. Determine the approach to enabling access to the frameworks and tools for business domains, developers and product teams
c. Start small, prove, and refine.
The topic will be presented and analysed by the experts at the Banking 4.0 Summer School, a program inspired by MBA business case methodology, June 10-13, Sinaia – Romania.
“This program was born from the belief that we need more critical thinking and applied learning in fintech. Not just understanding trends, but pressure-testing ideas, debating trade-offs, and simulating boardroom-level decisions. If you’re serious about shaping the future of finance, we’ve built this summer school for you.” – Nic Balaceanu, Banking 4.0 Summer School Program Director.
Banking 4.0 – „how was the experience for you”
„To be honest I think that Sinaia, your conference, is much better then Davos.”
Many more interesting quotes in the video below:
