The Fintech Power 50 Annual 2025

How autonomous agents will reshape financial services

Agentic AI is transitioning from concept to capability in retail banking, reshaping both how institutions operate and how customers interact. Unlike earlier AI systems that wait for prompts, autonomous agents perceive context, reason across multiple systems, act on behalf of users and learn from outcomes. All of this can be done instantly and at scale. This unlocks multiple benefits, including lower unit costs through full automation and higher revenue by providing proactive, personalised experiences. Early adopters are already shortening cycle times in onboarding and lending, increasing cross-sell through relevant recommendations and decreasing churn with anticipatory service solutions. The dawn of autonomous banking Agentic AI marks a significant step beyond traditional automation and even generative AI. These advanced systems blend machine learning, large language models and enterprise-wide automation to act as virtual coworkers operating 24/7. Major financial institutions like Bank of America and Citigroup believe that agentic AI could ignite a ‘corporate efficiency revolution’ with economic impacts that might exceed the internet’s influence on commerce. The distinction is important because chatbots follow scripts, while traditional AI analyses patterns. In contrast, agentic AI exhibits true perception, reasoning and self-learning abilities. This technology leads to what Citibank calls the ‘Do It For Me’ economy, where intelligent agents

and transaction processing. Automated KYC and AML processes with autonomous decision-making capabilities are decreasing human error while freeing employees for strategic work that requires human judgment and creativity. However, the technology’s most important impact will be seen in customer- facing applications. Agentic AI actively develops personalised financial solutions by understanding context, preferences and real-time needs. For example, when a customer transfers funds from savings to pay off debt, an AI agent might recognise underlying financial stress, assess their overall situation and instantly negotiate payment deferrals or restructure debt. This capability shifts banking from reactive service to proactive financial partnership. Implementation complexities Despite significant benefits, major challenges still remain. Infrastructure issues include connecting legacy systems, building unified customer data platforms and updating API architecture to support real-time agent interactions. These technical hurdles explain why, as Market Insights strategist Jim Perry observed, “it’s really only the largest banks doing any kind of significant experimentation” with agentic AI. Regulatory and ethical considerations also pose challenges. Banks must ensure algorithmic transparency, prevent bias in autonomous lending decisions, protect

data privacy in automated environments, and establish clear accountability frameworks for AI-driven financial decisions. Adding to the challenges, technological progress has outpaced regulatory guidance, forcing financial institutions to proactively collaborate with regulators while establishing strong governance frameworks. The competitive imperative Early adopters will gain significant first-mover advantages through superior customer experiences, lower operational costs, improved risk management and faster innovation capabilities. Success requires a strong vision and executive support across the organisation: n Set an institution-wide agentic AI ‘North Star’ tied to concrete P&L and customer metrics. n Prioritise two or three journeys (onboarding, servicing, lending) for full agentic re-design. n Invest in data plumbing first: real-time access, identity resolution, and event streams. n Stand up a cross-functional governance cell spanning risk, compliance, security and CX to approve agent behaviours and escalation thresholds. n Engage supervisors early with transparent model documentation and human-in-the-loop controls. n Upskill teams – pair AI engineers with product, operations, and risk owners to industrialise wins.

Jim Marous has been named as a top five influencer in banking for 10 straight years, an internationally known keynote speaker, co-

manage complexity, enabling employees and customers to focus on higher-value activities. Operational excellence The first wave of agentic AI deployment is already providing measurable benefits in back-office operations. Banks adopting these systems report significant improvements in compliance monitoring, fraud detection,

publisher of The Financial Brand , owner and publisher of the Digital Banking Report and host of the top retail banking podcast, Banking Transformed. Website: https://jimmarous.com