Half of the articles that appear in this magazine, the Spring 2026 edition of Financial IT, are devoted in some way to agentic Artificial Intelligence (AI).

 

These articles otherwise cover a huge variety of topics. Among others, these topics include payments, fraud detection, boosting the lifetime value of banking customers, modernisation of legacy systems and the strategic challenges posed by financial institutions’ adoption of AI.

 

This implies that, for now, there is no single and universally accepted definition of agentic AI. For now, though, we concur with the view of one of our contributors that agentic banking is ‘where AI doesn’t just respond, it executes. Autonomous agents can authenticate users, retrieve context, initiate workflows, orchestrate decisions across systems and complete tasks within defined compliance guardrails.’

 

Making all this work in practice will undoubtedly be difficult. As another of our contributors notes: ‘There is a major trust gap holding organisations back from scaling agentic AI projects beyond pilots. It happens time and time again – pressured by market forces and customer expectations, organisations are rushing into AI adoption only to find that for every new opportunity created, they must also plug new weaknesses that black-box algorithms have exposed.’

 

The frequent absence of trust is an important challenge. Our cover story explains how, in agentic banking, ‘the winners will not be those who deploy the most sophisticated models. They will be those who combine human expertise with an agentic workforce grounded in clean data, governed workflows, and transparent decisioning. In that three-layer model – human workforce, agentic workforce, and underlying systems – the banker remains the steward of the client relationship. The agent becomes the operator of complexity. And trust – earned through governance and orchestration - becomes the ultimate differentiator.’

 

In other words, successful deployment of agentic AI depends on a combination of factors, not least of which is the strategic vision of management. As a contributor to this edition of Financial IT points out, ‘the rise of agentic banking isn't simply about what AI can do for consumers, it's about whether financial institutions can transform themselves quickly enough to support that future. The infrastructure gap between current banking systems and the requirements of agentic finance is substantial, and it won't close overnight.’

 

Bearing this in mind, it is perhaps unsurprising that, so far, there are few concrete examples of positive outcomes from agentic AI. Nevertheless, one of our contributors does explain how ‘ a major European bank needed to modernize its mortgage processing system, a mission-critical application running on a 30-year-old mainframe. Traditional approaches projected an 18–24-month timeline and costs exceeding $50 million. By using AI powered assessment tools, the bank created a comprehensive application map in just six weeks, uncovering more than 2,300 hidden dependencies missed by manual analysis. AI generated documentation also surfaced critical business logic embedded in legacy code and undocumented for years.’

 

‘The bank then applied intelligent refactoring tools to break the monolithic system into microservices, with AI recommending service boundaries based on data flows and transaction volumes. Automated testing tools generated robust test suites for the new architecture, accelerating validation and reducing risk. The project was completed in 11 months at roughly 40% of the original cost, while the modernized platform now processes mortgage applications three times faster and cuts infrastructure costs by 60%.’

 

Over the remainder of 2026, it will become clearer exactly what agentic AI involves. By the end of the year, it should be a lot nearer than it is today to the mainstream of bank operations. Over the next 10 months, there will some high profile wins - where the revenue boost and/or cost reduction through adoption of agentic AI can be quantified. There will also likely be some (low profile) loss situations, where agentic AI has thrown up new problems. One absolute certainty is that interest in agentic AI will continue to expand.

 

The quest for knowledge about agentic AI - and much else at the intersection of technology and financial services - means that industry conferences should continue to flourish. Financial IT is pleased to be an industry partner of Money20/20 Asia in Bangkok (on 21-23 April), MPE in Berlin (on 17-19 March) and Pay360 in London (on 25-26 March). In relation to the last of these, we salute The Payments Association for the surge in the number of participants from 2,500 in 2024 to 6,000 this year.

 

We wish well to all the participants in each of these three important conferences - where there will be valuable insights to be gained about agentic AI. Meanwhile, we trust that this edition of Financial IT serves as a helpful guide to a rapidly emerging topic.