- Compliance & Regulation, Fraud, Risk, Technology
Share
Advanced artificial intelligence such as generative and agentic AI has been moving rapidly beyond the experimental stage and into the core of banking. In the process, it is providing an opportunity for institutions to rethink not just technology but how work is structured, how decisions are made, and what skills and staff they need to compete. What began as a push for efficiency is expanding into broader questions about governance, workforce strategy, and growth.
The 2026 ProSight Banking Outlook Survey highlights how quickly that shift is taking hold. Of the respondents with more than $10 billion in assets, all are already using AI or plan to do so within the next 12 months. Among banks under $10 billion, 84% said the same.
Isio Nelson, managing director of research, fraud, and thought leadership at ProSight, said the speed and breadth of adoption is striking. “It’s no longer a future state discussion,” he said. “We’ve seen a lot of shifts [in banking], but I don’t know that we’ve seen something so profound moving so fast—and with the potential for so much impact.”
ProSight recently gathered Nelson and a diverse panel of industry leaders to discuss how banks are rethinking the definition of roles, which capabilities matter, and how leadership teams can best guide organizations through a period of rapid AI advances.
This article features highlights from a conversation that included:
Where AI Is Already Changing the Work
While front-office applications of AI are still evolving, much of the early impact is showing up elsewhere in the bank.
“It’s remarkable how much AI has penetrated the middle and back office,” said Schuermann. “In areas including anti-money-laundering, the gains are already measurable. Banks are seeing a 30–50% reduction in customer due diligence [CDD] review volumes and a 60–80% reduction in false positives in transaction monitoring. Beyond efficiency gains, institutions are also reporting meaningful improvements in quality—identifying higher-risk clients more effectively in CDD and finding more instances of money laundering incidents,” Schuermann said. “It is really quite stunning.”
At 1st Source Bank, a $9.1 billion institution based in South Bend, Indiana, some early efforts have been focused on internal data and operations: streamlining workflows, improving efficiency, and building a foundation for broader use of AI, Short said. “One of the first places we’re starting is operational efficiencies in the back office, whether that’s in credit and deposit services, loan operations, or wherever they may be,” she said.
But for Short and many other thought leaders, the more consequential shift from AI will not be about doing existing tasks faster through automation. Rather, it will involve revisiting assumptions about how work is performed and who—or what, as the case may be—performs it. “We are really rewriting in very basic ways how work gets done,” LaMonica said.
Financial institution leaders envision reconstructing roles around higher-value activities, LaMonica said. In compliance and operations, that may mean shifting time away from manual review toward investigation, escalation, and judgment.
The Changes to Come
While early advanced AI efforts focus on cost, “the longer-term opportunity is on the revenue side,” LaMonica said. “That’s where it gets really exciting.”
At 1st Source, Short said, in addition to the drive for efficiency with AI, they are also striving to “enhance the customer and client experience.”
Nelson said AI assists can help bankers to “better know their customers, be able to research them before they have their conversations, and have more information”—all from one convenient platform.
Far from the worry of making banking impersonal, Schuermann said, with “AI supporting and helping the banker, those human-to-human interactions [with customers] can be made much richer and more personalized.” That phenomenon “leans against the perception of what a lot of AI does,” he said.
In addition to making conversations with customers more meaningful, Short said, AI can also free up more time for relationship managers to have those conversations. She said banks are examining employees’ workflow to determine which processes or steps AI can eliminate or shorten, and then asking, “How many hours does that free up?”
Making the most of AI, of course, will require employees to be comfortable and conversant with it. At 1st Source, teams are being pushed to engage with new tools and explore how AI can be applied across different parts of the business. Short said that mindset is becoming as important as technical skill.
“We have to have people who are curious,” Short said. Meanwhile, leaders should give employees room to experiment, recognizing that not every effort will be tightly focused at the outset as the bank works through a range of potential use cases.
Short described one case where an employee with strong Excel skills sat with a colleague using Copilot and began exploring how the technology could improve her own workflows. The goal, Short said, is to help employees “get comfortable with AI” by building on capabilities they already have.
Employee Buy-In and Leadership’s Role
One element of AI that can encourage employees to dive in is “it’s not something where you have to be technically savvy to be able to use it,” Nelson said. “You can apply your knowledge of what you’re trying to get done and use AI to assist or augment what you are doing.”
For employee buy-in, it’s also important that leaders demonstrate that they are using and benefitting from AI, Hsu said. “If the only pronouncement is ‘We’re going to adopt AI,’ the natural interpretation is, ‘I’m going to get fired,’” Hsu said.
On the other hand, he said, “if leaders put hands on keys and play around with this stuff and talk about the opportunity side, it takes some of the pressure off on the efficiency side. Leaders can’t just delegate it and say, ‘Produce great things.’”
LaMonica put it this way: “All roads come back to leaders and leadership. So much of the success depends on leadership culture.”
Going forward into the AI era, Hsu expects a shift in banking to roles such as “integrators, builders, and trust engineers,” reflecting the need to connect AI systems to business context, develop new capabilities, and ensure those systems operate safely at scale.
There will be seismic shifts in how regulators do their jobs as well, Hsu said. “Regulators like frameworks: First, line, second line, and third line people. Processes and systems,” he said. “Frameworks are helpful because they tend to align with good controls, risk management, and remediation. With AI, I think these lines are going to get blurred. Where is the division between the first, second, and third lines when you’ve got AI agents [in all three areas]?
As those distinctions blur, traditional models of accountability become harder to apply. Hsu said oversight is likely to place greater emphasis on benchmarking and structured ways of testing systems—such as independent evaluations and scenario-based challenges—rather than relying primarily on whether prescribed processes have been followed.
“Because whatever process that is today, it’s going to change when there is a new model release,” he said.
As AI becomes more embedded, Hsu said, financial institutions should not “look to regulators as to where the guardrails and the guidelines on this should be. The approach in the U.S. today is to let the growth happen.” While that realization can be daunting, he said, it’s also “a fantastic opportunity for the industry to say, ‘what does good look like?’” in terms of compliance. “It’s not going to be artificially imposed by current regulators.”
Throughout the financial industry, LaMonica said, this new dynamic will require “people who can adapt and be flexible, who can change quickly, who thrive on change, and can learn quickly,” she said.
At Providence, Rhode Island-based Citizens, she said, the $227 billion bank is “incorporating that into the profile of the kind of person we’re looking for across the organization.”
Other important characteristics for effectiveness in an AI-enabled workplace, the leaders said, include curiosity, good judgment, ethical reasoning, client advisory skills, and critical thinking.
Closing Thoughts
As the financial industry anticipates and works toward a transformation in how the very work of banking gets done—and the historic opportunities this will create—it also faces uncertainty and new risks.
As AI becomes more embedded in core workflows, Schuermann said, errors are no longer isolated or slow-moving—they can propagate quickly across systems and processes. “With technologies that improve things at speed and scale, it means that when something goes wrong, it will go wrong at speed and at scale,” he said.
Another possible factor limiting efficiency gains might be the current cost of the work being automated. Some of the functions most immediately affected by AI, including reconciliation and model risk management activities, are already offshored or lower-cost, Schuermann said. That has implications for expected savings. “If this technology is mostly good at replacing already lower-cost labor, then the cost savings aren’t going to be quite as high as one might have hoped,” he said.
As financial institutions work to address these and other questions and challenges in deploying AI widely, Nelson said, they will “have to learn from each other—both what works and what doesn’t—along the way so the industry grows as a whole.”
Peer sharing events, including those offered by ProSight, have traditionally provided venues for the industry to share and discuss practices on critical developments and topics. Nelson said organizations might also consider standing up their own work groups—spanning functions and seniority—to best explore the possibilities of human + AI approaches.
They may find benefit from looking completely outside financial services, too. AI “is a general purpose technology and has the potential to change so much about how we work,” Schuermann said. “We’ll learn a few things by looking at other industries to see what they’re doing.”
No matter where inspiration is taken, Schuermann said, “this is a time for a mindset that is very open to safe exploration and change. Exploration and experimentation is absolutely vital.”
*This discussion occurred prior to the issuance of SR letter 26-2, Revised Guidance on Model Risk Management, by the Board of Governors of the Federal Reserve System, Office of the Comptroller of the Currency, and Federal Deposit Insurance Corporation, which replaced SR letter 11-7, Guidance on Model Risk Management.
Join our community to unlock exclusive content, connect with industry experts, and gain access to valuable resources that will help you stay ahead.