Artificial Intelligence is the most transformative technology of our time, and its impact on financial services is inevitable. Yet despite the hype surrounding GenAI, most financial institutions are still in the experimental phase. The bigger challenge? Many are thinking about how to adopt AI, but not yet about how to adapt to it.
Adopt AI vs. Adapt to AI: The Two-Sided Challenge for Financial Institutions
As we move into an agentic future, AI presents a dual challenge for FIs:
- Adopting AI (agents) – How can FIs integrate AI agents into their operations, products and services?
- Adapting to AI (agents)– How will FIs respond when consumers and businesses begin using AI agents to interact with them?
GenAI: Slow Start, Big Finish in Financial Services
Financial services companies have been leveraging AI/ML for decades—think JPM, Citi, Cap1, Visa, Amex—especially in areas like risk and fraud. But when it comes to GenAI, adoption has been more cautious.
So far, the go-to entry point has been through Microsoft Copilot or name-your-foundation-model enterprise version for employee productivity. These tools are easy to roll out—they’re built into enterprise software that FIs already trust. In contrast, partnering with early-stage GenAI startups, especially for critical or customer-facing tasks, is harder to swallow.
Despite the hype and experimentation, production use of GenAI in high-risk or customer-facing applications is still rare. Why? Security, compliance, accuracy, and control. FIs can’t afford even small errors.
That said, momentum is building. Many FIs are hiring Heads of AI, setting up internal AI committees (which, yes, adds yet another “manual” layer to already painful onboarding process), and beginning to partner with AI-native startups. Back- and middle-office functions are proving to be fertile ground. As new trust frameworks evolve and guardrails improve, expect adoption to accelerate rapidly. After all, FIs may move slowly—but once the risk is manageable, no one scales like they do.
Among the biggest opportunities? Compliance automation. Big banks spend millions on armies of compliance staff. Others can’t hire fast enough. New autonomous agents can now complete work that once took weeks in a few seconds.
AI Agents Are Coming—Will Banks Be Ready?
“Know Your Agent” (KYA) is the new KYC.
Soon, consumers won’t just log into banking apps—they’ll send AI agents to do it for them. These bots will manage finances, shop, and interact with businesses on users’ behalf. In this world, banks must serve not just customers, but customer-deployed bots.
The challenge? How does a bank differentiate between a legitimate AI agent and a malicious bot?
Enter KYA—Know Your Agent. We believe that just like KYC or KYB, banks will need to onboard and authenticate AI agents with unique credentials, ensuring they’re legit and acting with permission.
Some of the key challenges in this new landscape remain the same as today, only in the form of agents, and new ones emerge:
- Account Takeover (ATO): Is the AI agent really working for the user—or has it been hijacked?
- New Account Fraud: Is the new customer a fraudulent agent or a legitimate agent?
The first phase of agentic banking will be Agent-to-Business (A2B) Interactions (“Bot-to-Bank”), where banks will need to implement solutions to validate and authorize AI-driven transactions while preventing abuse. The next phase will be Agent-to-Agent (A2a)) interactions in which the customer agent interacts directly with the banks AI agents.
As AI agents become the new intermediaries, fraud will evolve as well. GenAI-powered fraudsters will automate attacks, making fraud economically viable at an unprecedented scale. AI-driven security solutions must keep up, using cryptographic handshakes, behavioral analysis, and anomaly detection to distinguish between good and bad actors.
Bottom line: AI agents aren’t just changing how customers bank—they’re redefining who the “customer” is. FIs will have to keep up, or risk getting outsmarted by the bots.
