Beyond Efficiency: How Human-in-the-Loop AI Is Redefining the Contact Center
By Nicole Volpe, Contributor at The Financial Brand
Simple Subscribe
Subscribe Now!
Executive Summary
- AI-powered fraud detection in contact centers now catches behavioral anomalies humans miss, analyzing vocal inflection and phrasing patterns while reducing compliance costs from the $61 billion banks spend annually on financial-crime compliance.
- Human agents need new support systems as AI automation removes routine tasks, leaving only complex, emotionally charged interactions. Forward-thinking organizations now employ licensed clinicians to prevent burnout among contact center staff.
- Contact centers are becoming continuous learning engines where AI systems capture agent knowledge, build institutional memory, and surface relevant information in real time — transforming these operations from cost centers into strategic assets.
Banks and credit unions, like most businesses today, are in the hunt for AI upsides — ways in which the new technology can speed operations and reduce expenses without eroding quality or trust. Under pressure from boards and shareholders, institutions are looking for high-potential targets that might yield efficiency gains, including especially big cost centers that are heavily dependent on both people and processing.
They’re also increasingly wary. A much-discussed study released by MIT this year revealed that 95% of corporate generative AI pilots are failing to deliver measurable business impact or return on investment. And in banking, a new analysis of large institutions by Evident, which tracks AI adoption in financial services, revealed that roughly 70% have nothing tangible to show for their AI investments.
But statistics about AI projects failing to hit P&L and ROI goals only tell part of the story — and in fact may overlook ways in which AI can add critical value to an institution’s overall business, even without accruing immediate bottom-line gains.
Contact center operations make for a compelling case study in this regard. At first glance, these labor- and transaction-intensive operations appear to be exactly the sort of ripe expense reduction target P&L hawks are scouting for. A closer look, however, suggests that greater advantages might come from unlocking new and different categories of upside.
Strategic Touchpoints
Contact center work is deceptively simple, simultaneously high-touch and high-throughput. These operations rely heavily on leveraging customer data, yet most still struggle to navigate the multiple siloed databases they must tap into to excel.
Against this backdrop, with little fanfare, forward-thinking organizations are increasingly viewing contact centers as strategic touchpoints. They’re exploring new use cases that can drive brand, engagement, and sales, and increasingly incorporating AI into their efforts.
“Let’s start with the basic idea that everyone wants their issues solved quicker,” said Chris DeLambo, Division Vice President of Agentic AI Solutions at TaskUs, a global provider of outsourced digital services and customer experience. “Faster resolutions lead to happier customers and members, but they also lead to happier employees — so we are improving both customer lifetime value and employee retention.”
Elevating people in human-in-the-loop processes is a keystone concept. Those that do it right are enhancing employee well-being as much as they are member and accountholder outcomes. They’re optimizing data insights. And they’re fostering human/AI interactions that enable progressive learning. This article explores three under-recognized sources of AI-driven value within contact centers.
Good Cop, Better Cop
Regulatory compliance and fraud management are critical functions within contact center operations that are both labor-intensive and fraught with downside risk. With regulatory expectations growing and the cost of maintaining compliant operations climbing, financial institutions know they have room to improve.
Institutions in the U.S. and Canada spend about $61 billion a year on financial-crime compliance, according to a 2024 survey by Lexis-Nexis and Forrester. In the survey, small, midsize, and large institutions reported that their own compliance costs are rising, and uniformly cited automating compliance work as their top pain point.
At the same time, front-line service channels have become prime targets for sophisticated attacks, from vishing and deepfake voice fraud to social-engineering attempts designed to trick representatives. Roughly 6% of inbound calls to call centers were considered high-risk for fraud in 2024, more than double the rate in 2022.
AI systems can have an outsized impact — both reducing overall compliance work and improving fraud prevention. AI agents now monitor voice and digital channels, flagging fraud through tone and language patterns that humans can miss. These tools have already improved fraud detection for major banks by analyzing vocal inflection and phrasing for behavioral anomalies. This results in team members spending less time acting as investigators and more time providing quality service and solving customer problems.
More generally, AI systems ensure all required disclosures and legal phrasing are delivered verbatim and track 100% of conversations for audit, which is a significant advance from traditional approaches. “Not long ago human agents needed weeks of training to master scripts and protocols,” said TaskUs’s DeLambo, whose organization has deployed this capability on behalf of client institutions. “Now the system listens to the conversation and feeds them the right script in real time.”
Dig deeper:
- Banks Can Convert Messy Data into Unstoppable Growth
- Your Biggest Differentiator: Caring about Customers
- Why AI Won’t Replace Bankers but Will Expose the Bad Ones
Well-Being by Design
As customer expectations rise and budgets tighten, contact center reps bear a higher emotional load. While definitive data is hard to come by, turnover rates among contact center staff range from 30% to as high as 60%, across industries, with stress the leading driver. AI can help correct this, though not necessarily in predictable ways.
The first step is what many observers would expect: Let AI handle the repetitive questions that make up the bulk of contact-center work: capturing names, phone numbers, the reason for the call, etc. Removing these tasks allows human agents to focus on more complex, problem-solving interactions that draw on their knowledge and tools. These professionals can also focus on high-value or high-net-worth customers whose interactions require more judgment and personalization.
“For those agents that really like to solve problems and handle more complex work, it’s better work,” DeLambo said. “It draws on more of your knowledge, it’s more exciting, more interesting — it’s a better job in many ways.”
At the same time, though, it’s a more stressful job, with heightened risk of fatigue and burnout, because when routine calls are automated away, all that remains are the complex, emotionally charged conversations that demand patience, judgment, and empathy. In the past, workers had natural pauses to handle after-call tasks, which offered brief mental breaks. Now, even those moments are automated away.
According to DeLambo, TaskUs employs more than 200 licensed clinicians to support its frontline staff. As AI takes on more of the routine volume, structured mental-health support is becoming an essential part of sustaining human performance. “These professionals are working with our agents and our front line staff to help them with the mental health part of their work,” De Lambo said. “We see that as a key component as we get deeper into AI and deeper into these complex, emotionally charged conversations.”
The result is a less mind-numbing job, with greater opportunity to learn and grow, but new support systems are necessary to sustain it: more humans in the loop helping the reps achieve their potential.
A Learning Engine
Customer-service operations are knowledge-intensive by nature, yet much of that knowledge remains trapped in silos. Every day, agents answer questions, solve exceptions, and interpret policy in ways that are not recorded. Complaint patterns take too long to identify and responses too long to spread across the organization. New products that might turn a basic T&Cs inquiry into an upsell opportunity are never offered.
AI is beginning to close these gaps. For starters, it can give reps’ touchless access to their institution’s knowledge base — e.g., product information, training modules, downtime alerts — by listening for subject-matter cues and surfacing relevant information in real time while the agent stays focused on the conversation. Even more compelling, though, is the potential for AI-enabled continuous learning systems.
Here’s how it might work in practice. When a customer asks a question the AI system can’t answer, the system might route the query to a human agent. The agent resolves the issue, and that interaction is logged for review. The audit team later reviews the issue, confirms the correct response, and adds it to the knowledge base. Over time, this human feedback loop makes the system smarter, reducing future handoffs and improving accuracy for the next customer.
In implementing such systems, it’s not sufficient to install AI as a straightforward read/write interface to the knowledge base. “The smarter approach is building defined workflow processes first, so the AI is drawing on a knowledge base that you’ve built for purpose and refined,” DeLambo said. “Know the businesses you’re trying to automate. If you don’t have the right knowledge base, you’ll end up with hallucinations.”
A Bigger Opportunity
The contact center is fast emerging as one of the industry’s most forward-leaning touchpoints — a far cry from the non-mission-critical, back-office function it once was. Institutions looking to make the leap from AI efficiency to AI value creation could do worse than give these operating units a closer look, not least because they offer self-contained test cases that give AI-cautious institutions comfort as they wade into uncharted waters.
Yet the opportunity is bigger than that. Many of today’s most insightful thinkers and organizational strategists see AI’s greatest potential in its pairing with human talent. And contact centers, with their large customer-facing workforces, offer fertile ground to cultivate that partnership.
Employees are being asked to master new and sometimes disorienting tools and workflows, all while maintaining their usual daily performance. While this may lead institutions to see temporary productivity gains, it risks burning out their most valuable people. In truth, they must look well beyond short-term metrics: as these front-line teams learn to work alongside AI, they are actively engaging their uniquely human capabilities. In doing so, they help build new organizational “muscle memory” — ingrained habits and shared knowledge — that strengthens collaboration and effectiveness between both people and machines.
