How Financial Institutions Are Leaning into AI to Orchestrate the Customer Experience
By David Longbardi
Simple Subscribe
Subscribe Now!
As early AI deployments start to show tangible gains, many of the financial industry’s most promising applications are emerging in customer experience, where high volumes of customer-interaction data, a significant concentration of operating expense, and a direct connection to revenue combine to shape powerful use cases.
These advances come as more financial service providers, from fintechs to community banks and credit unions, push past their initial AI hesitancy. Perhaps the most telling finding in a spring 2026 American Bankers Association survey was that financial institutions ranked AI as their most important issue to address, ahead of cybersecurity and data privacy — and identified “doing nothing” as the greatest AI-related risk of all.
Key insight: Customer experience offers a particularly strong case for moving forward. Automating routine service interactions can lower operating costs, while faster resolutions and more personalized engagement can improve satisfaction, retention, and conversion. “For the first time in a generation, the industry has a real shot at truly transforming customer experience,” said Hardy Myers, vice president of global strategic partnerships at NiCE, whose NiCE Cognigy unit helps financial institutions design and orchestrate AI.
Myers cites the leverage institutions gain when they progress from generative to agentic AI, and from there to orchestrated networks of agents. GenAI produces content or answers questions in response to user prompts or contextual inputs. Agents are given instructions and access to systems of record so they can interact with LLMs to carry out defined tasks on an accountholder’s behalf. Orchestration extends that capability by coordinating multiple agents and technologies in the service of a broader objective.
“You’re not telling them how to do it, you’re telling them what to do,” Myers said. The institution’s implementation team defines the larger objective — e.g., advance a loan prospect to a booked appointment with a lending specialist — while the agent determines, within set parameters, how to accomplish it.
Drawing on live client implementations, Myers offered perspective into how customer experience automation can progress: from single-channel GenAI applications, to human-and-AI interactions orchestrated across multiple channels, to fully agentic systems that can take action and carry customer journeys forward.
Case Study 1: Automation Plus Live Handoff
A major global consumer bank with more than 180 million customers turned to AI to solve a basic lending problem: too much advisor time was being spent trying to reach customers, including qualified prospects, who were not ready or available to talk.
The bank had a pool of customers who had previously submitted loan applications but had never received a final response. It wanted to reconnect with those customers, offer them appropriate alternative options, and complete eligible loan contracts by phone. But loan advisors were spending substantial time on failed call attempts, often reaching voicemail or the wrong person.
To address that problem, the bank has deployed an AI agent to handle the initial outreach. The agent contacts interested customers, filters out those who are unresponsive or no longer interested, and reschedules calls when the timing is wrong. Once a customer engages, the agent can transfer the call directly to a loan advisor or schedule an appointment for a later time.
Key insight: The value lies in separating the repetitive work of reaching customers from the higher-value work advisors do. Rather than having loan officers spend time dialing unanswered numbers and leaving messages, the AI agent keeps working through the contact list until it identifies customers who are ready to continue the conversation.
According to Myers, the agent achieved an 80% successful handover rate among customers it reached and filtered out 85% of unsuccessful contact attempts before they consumed an advisor’s time. The bank moved from planning to go-live in three months.
Case Study 2: Systematically Pursuing a Larger Goal at Fairstone Bank
Fairstone Bank offers an example of agentic AI orchestration. The Canadian lender first tried to improve follow-up on loan prospects by centralizing the process across its 250-plus branches. That brought more consistency and increased the number of prospects entering the funnel, but loan-booking rates continued to decline. Fairstone needed a way to pursue those opportunities more persistently and systematically without simply adding more manual follow-up.
To do this, the bank deployed a set of AI agents that would manage a two-week, multichannel sequence of digital interactions with loan prospects. Working together, the agents sustain engagement over multiple touches, tailoring conversations to different customer segments and answering questions — all with the larger goal of moving prospects to schedule an appointment with a lending specialist. The larger orchestration model applies customer consent and contact-frequency regulations, and respects channel preferences, including movement from email to SMS when appropriate.
That ability to pursue an objective across time is what distinguishes the application from the conversational AI used by hummgroup and Rentenbank. The system works toward an outcome — re-engaging prospects and advancing their journey. And it adapts along the way as needed, including by arranging callbacks when the timing is wrong.
Key takeaway: Fairstone says 65% of prospects responded to the outreach and, among those who did, 90% booked appointments. Within four months, loan bookings were up 10%, well above the bank’s original 4% target, while the sales cycle was 7% shorter. The bank also reports millions of dollars in annual net new cash and an ROI of more than 20-to-1.
A Different Kind of Transformation
The larger lesson of these initiatives is AI’s potential to change how financial institutions think about technology-driven transformation. Transformations have a reputation for requiring large upfront investments, multiyear timelines, and long stretches in which institutions see little in the way of results. (Research from Bain & Company indicates that 88% of business transformations fail to fully achieve their original strategic ambitions.)
Myers argues that AI can change that equation. Unlike typical technology transformations, AI implementations, under the right circumstances, can run on a fast feedback loop. Rather than require years of investment before yielding results, AI applications can fail, or succeed, fast — enabling continuous course correction and fewer costly dead ends.
These characteristics can combine to create a flywheel effect, in Myers’ view. Once an institution starts to achieve results with AI, its efficiencies or cashflow, can fund the next AI initiative. Fairstone Bank’s lending application is a case in point, he said: It paid for itself in the short term and the bank has since decided to extend the approach into French-language experiences, as well as credit card marketing and cross-selling applications.
