How AI Is Changing the Contact Center Agent's Job
What human agents do when AI handles the routine work, and the new roles taking shape around them
Make CX Current News one of your go-to sources on Google
AI is taking over a growing share of routine contact center work, but most organizations are keeping their human agents and changing what those agents do. Agents now spend more of their time on complex, sensitive and high-value conversations, and new roles are forming around supervising, building and evaluating AI agents.
What is happening to contact center headcount
A few years ago, many forecasts assumed AI would shrink contact center staffing quickly. Organizations that have actually deployed it report a slower and more uneven change. In a March 2025 Gartner poll of 163 customer service leaders, 95% said they planned to keep human agents, and Gartner predicts that by 2027, half of the organizations that expected to significantly cut their customer service workforce will abandon those plans.
The firm went further a few months later, predicting that no Fortune 500 company will have fully eliminated human customer service by 2028. Kathy Ross, a senior director analyst in Gartner's customer service practice, said, "We expect fewer human agents, but not completely agentless organizations."
Cresta, which publishes CX Current, surveyed 300 CX, support, operations and IT leaders for its 2026 CX Workforce Report. Respondents said only 9% of their conversations were handled entirely by AI, and 97% said AI had allowed them to move staff into higher-value work.
What AI is taking off the frontline agent's plate
AI usually takes on the most repetitive requests first, such as password resets, order status checks, appointment changes, balance inquiries and other requests with a clear answer and a short path to resolution. AI agents now handle many of these conversations from start to finish. When they can't, a well-designed handoff passes the full context to a person so the customer doesn't have to start over. Brinks Home built its AI agent to transfer calls with a complete summary of the conversation so far, which spares both the customer and the human agent from repeating the basics.
At Brinks Home, Veronica Moturi, then the company's SVP of customer experience, said in early 2025 that "our agents love working with it," crediting the AI agent with smoother transfers when a call needs to escalate. Cresta reported that Brinks agents particularly valued no longer having to handle routine steps such as authentication and call summaries.
For the conversations that still reach a person, AI is absorbing much of the administrative work around the call. Real-time assistants surface knowledge and suggested responses during the conversation, and automatic summaries replace most after-call note-taking. Cresta's Knowledge Agent, for example, listens to the live call and reads customer details on the CRM screen to bring up the policy that applies without a manual search.
A Stanford and MIT study of 5,179 support agents, published by the National Bureau of Economic Research, found a 14% average increase in issues resolved per hour, with the largest gains among newer and less experienced agents. The researchers also found improvements in customer sentiment and employee retention.
What stays with people
Ninety-three percent of the leaders in Cresta's survey said the calls their human agents handle are becoming more complex, with a larger share of exceptions, billing disputes, emotionally charged situations and high-value sales or retention conversations. These are the interactions where mistakes cost the most and where customers most want a person. When Gartner surveyed more than 5,700 customers about AI in customer service, difficulty reaching a person topped their list of concerns.
Asked which human skills mattered most, the leaders in Cresta's survey named emotional intelligence, empathy, judgment, decision-making in complex situations, critical thinking and problem-solving. Speed and script adherence, long the core of agent scorecards, count for less when the remaining calls rarely follow a script.
Devon Mychal, Cresta's VP of product marketing, has described the purpose of AI assistance in similar terms. Taking complicated system work off an agent's plate, he said in a discussion with Metrigy, leaves people free to "connect with one another and listen and show empathy."
A shift made up almost entirely of escalations and disputes is harder on an agent than one with a mix of simple and difficult calls, and contact centers will need to reflect that in scheduling, breaks and the support they give frontline staff.
New roles forming around AI agents
Running AI agents in production creates work that didn't exist a few years ago, and much of it draws on frontline experience. One recommendation in Cresta's workforce report is that organizations formalize this new AI work instead of leaving it as an informal side duty, and three roles in particular are taking shape.
AI supervisor
Cresta introduced the AI supervisor role in December 2025 alongside its Agent Operations Center. An AI supervisor monitors live AI agent conversations, gets alerted when one needs help and can either guide the AI agent or take over the conversation without the customer having to repeat anything. The Cresta designers who built the tool describe the job as designing systems of supervision and trust for the moments when an AI agent stumbles. Experienced agents and team leads are strong candidates because they already know what a good resolution looks like.
AI agent builder
Someone has to turn standard operating procedures and real customer conversations into an agent's instructions, workflows and guardrails, and then keep refining them as products and policies change. Cresta's applied research team describes AI agent development as a cycle of scoping from historical conversations, simulating realistic customers, testing and improving after launch. People with deep knowledge of how customers actually describe their problems are valuable in that cycle, whether or not they come from a technical background.
AI quality analyst
Quality assurance work moves from listening to a sample of calls toward writing requirements for automated graders, checking those graders against expert judgment and investigating the conversations they flag. Cresta's engineering team calibrates its automated evaluators against human reviewers until their verdicts match, and that calibration depends on people who know the business rules well.
How training and coaching change
New hires have traditionally learned on the easy calls. When AI handles those, new agents meet harder conversations much sooner, with less room to build confidence first, which makes preparation before the first live call more important.
Scripted role-play with a trainer is hard to scale and seldom reflects how real customers behave. Cresta's Training Simulator, launched in July 2026, uses AI agents to play simulated customers modeled on a company's actual conversations and grades each practice session with the same quality criteria applied to live calls. Managers can assign scenarios tied to the specific behaviors an individual agent needs to work on.
Cresta CEO Ping Wu said at launch that the aim is for every agent to be able to practice a hard conversation "anytime, as many times as it takes to get it right." A new agent's early mistakes then happen in practice, where they don't cost a real customer anything.
When every conversation can be scored automatically, a supervisor can coach from patterns across hundreds of an agent's calls instead of a handful of samples. The NBER researchers also found evidence that AI assistance helped spread the practices of the most skilled agents to newer ones, which suggests that well-designed AI can shorten the time it takes a new agent to reach full proficiency.
What CX leaders should do now
Plan for a different mix of work. Map which conversation types are moving to AI, which will stay with people and what skills the remaining work requires, then build hiring and staffing plans from that map.
Rewrite agent scorecards for complex work. Give resolution quality, judgment and customer outcomes more weight than handle time and script adherence.
Create formal AI roles with career paths. AI supervisors, builders and quality analysts need job descriptions, training and a clear route for frontline agents to move into them.
Train before the first live call. New agents will face harder conversations sooner, so realistic practice needs to come earlier in onboarding.
Fix data access and integrations. In Cresta's survey, 81% of leaders cited integration complexity as the biggest barrier to AI adoption, and only 7% said they could easily access their own conversation data.
Watch frontline wellbeing as the call mix gets harder, and adjust schedules and support before burnout shows up in attrition numbers.
Frequently asked questions
Will AI replace contact center agents?
Current evidence points to fewer agents doing different work rather than full replacement. Gartner predicts that no Fortune 500 company will have fully eliminated human customer service by 2028 and that half of the organizations planning major cuts will abandon them by 2027.
What will contact center agents do as AI takes on more work?
They will handle a larger share of complex, sensitive and high-value conversations. Some will move into roles that supervise, build and evaluate AI agents.
What skills do contact center agents need when AI handles routine work?
Leaders surveyed by Cresta ranked emotional intelligence, empathy, judgment, decision-making in complex situations, critical thinking and problem-solving as the most important.
What is an AI supervisor in a contact center?
An AI supervisor monitors AI agent conversations in real time and steps in to guide the AI agent or take over when a conversation needs human judgment.
How should contact center training change?
Because new agents encounter difficult conversations sooner, training needs more realistic practice before live calls, such as simulations built from a company's real customer conversations.
