Agentic AI offers real opportunities, but only if processes are redesigned, governance is strengthened, and talent development keeps pace with technological acceleration. Opinion of Sophia Palmstedt, Director, Advisory at Gartner, ahead of a presentation she will give on the topic in Amsterdam.
“Investment willingness is high,” she says. “But high investment and high impact are not automatically the same.” Recent research by Gartner shows that 17 percent of bank CIOs have already implemented AI agents. Another 41 percent say they plan to do so within the next twelve months. According to Palmstedt, this indicates that the technology is entering an acceleration phase.
At the same time, Gartner predicts that fewer than half of banks will achieve significant revenue growth from AI by 2029. “That does not mean AI agents do not deliver value,” she explains. “But it does mean that structural performance improvement requires more than simply adding technology.”
Where AI Agents Are Being Applied Today
When Palmstedt speaks with banking executives, two questions consistently arise. “The first is: where should we start? The second is: what could go wrong?” In terms of applications, she sees clear patterns. “Fraud detection, customer service, and IT service management are most frequently mentioned as starting points.” In addition, AI agents are being applied in lending, payment processing, and quantitative investment strategies.
The appeal lies in the ability to automate more complex workflows. “Multi-agent systems can coordinate multiple decision steps without requiring continuous human intervention. That is a fundamental difference from earlier generations of automation.” At the same time, she adds an important caveat. “If you automate an inefficient process, you end up with an inefficient automated process. AI does not fix structural design flaws.”
Process Redesign as a Prerequisite
According to Palmstedt, organizations often underestimate how frequently their existing processes are not ready for autonomous AI. “Many banking workflows have developed historically. They are shaped by compliance requirements, mergers, legacy systems, and organizational silos.” Adding agentic AI to such an environment requires reconsideration. “You have to ask yourself: if I were to redesign this process today with AI as the starting point, what would it look like?”
This goes beyond an IT project. “It affects operational design, risk management, and governance. It is a business transformation exercise, not just a technological one.” Palmstedt also emphasizes the importance of data quality. “We often hear: our data is good enough. But as soon as agents start acting autonomously, inconsistencies become visible.”
Banks therefore need to critically assess accessibility, data definitions, and ownership. “Autonomous systems require consistent and reliable information. Otherwise you increase risks instead of reducing them.”
Risk Management
Gartner research shows that three themes dominate discussions among banking CIOs: compliance, data privacy and security, and autonomous decision-making without human oversight. “Autonomy is attractive, but autonomy without supervision is problematic,” Palmstedt says.
She points to the importance of clear agent hierarchies. “You must explicitly define which agent has access to which systems and who ultimately holds decision responsibility.” She also observes a growing market for AI security platforms. “Organizations are looking for ways to allow agents to collaborate safely across silos.”
Another notable development is the use of AI to monitor AI. “We are seeing the emergence of so-called guardian agents. Their role is not to answer customer questions, but to supervise other agents.” According to Palmstedt, this is not a luxury. “If you scale autonomy, you must also scale oversight.”
What AI Agents Deliver
Banks already working with AI agents primarily report operational benefits. “Time savings and improved accuracy are frequently mentioned outcomes,” Palmstedt explains. She stresses that these are valuable indicators but not guarantees of strategic impact. “Efficiency matters, but revenue growth requires a broader playing field.”
CIOs should therefore remain critical when evaluating alternatives. “Sometimes traditional automation is more effective than an AI agent. Technology choices must be driven by use cases, not by hype.”
Talent Shortages
A recurring theme in discussions with banking executives is the availability of qualified talent. “Technology, skills, and talent were the second-largest barrier to AI implementation in banking in 2025,” Palmstedt says. She expects this issue to persist. “We are seeing accelerating adoption, but workforce transformation programs often lag behind.”
CIOs need to take an active role in addressing this challenge. “Implementing AI without structurally investing in skills development is not sustainable.” This requires not only technical training but also change management. “You need to guide people through how their roles are evolving. Otherwise you risk resistance or underutilization.”
Two Systems of Intelligence
At the executive level, a broader strategic question emerges: how do AI agents relate to the human workforce? Palmstedt describes this as “the power of two intelligence systems.”
“AI excels at analytical consistency and scalability. Humans excel at empathy, context, and moral judgment.”
The key, she argues, lies in complementarity. “Organizations that use AI to strengthen human capabilities create more value than those focusing solely on replacement.”
For CIOs, this means responsibility not only for technology but also for how technology collaborates with people. “The real differentiation lies in how well you organize that collaboration.”
Toward the Autonomous Enterprise
Globally, Gartner observes a wide range of agentic AI applications in banking. Yet the underlying trend remains consistent. “We are moving toward what we call the era of the autonomous business,” Palmstedt says.
This evolution includes an augmented workforce, autonomous operations, and products that adapt to changing customer needs. “Banks are accelerating that shift,” she notes. “But acceleration without direction does not create advantage. Experiment and invest, but ensure that your foundations, governance, and talent development grow at the same pace.”
