AI is absorbing routine contacts, but most contact centres are redesigning the agent role rather than cutting headcount.
Picture an agent who used to spend the morning on order-status checks and password resets. AI now handles most of those. The agent's queue is shorter, but every contact left in it is harder: a billing dispute, an angry customer, a policy question with no clean answer. The job did not disappear. It changed shape.
This article is about that change. What the redesigned agent job becomes, what skills it needs, and what it means for how you hire, train, and pay.
What the Gartner research found
Gartner surveyed customer service and support leaders and found the pressure to adopt AI is near universal. In a survey of 321 leaders, 91% said they are under pressure to implement AI in 2026. The interesting part is what they are doing with their people. A later Gartner survey found that 85% of service and support leaders are expanding human agent responsibilities, and just 31% have implemented or are planning frontline layoffs in response to AI through the first quarter of 2027.
So the headline expectation, mass layoffs, is not what most leaders are doing. Coverage of the research describes it plainly as workforce redesign. The same Gartner work found 75% of leaders are shifting agents into new roles, and 84% are revising the skills or experience they ask for in new hires.
What the redesigned job looks like
When AI takes the simple contacts, what is left for a person is the harder mix: complex problems, emotional conversations, policy-heavy cases. The agent moves from processing contacts quickly to handling fewer contacts that need judgment.
New tasks come with that shift. One example: Gartner found 58% of leaders want to move agents into knowledge management, keeping the content accurate that both the AI and customer self-service rely on. An agent who knows where the AI gets things wrong is well placed to fix the source. Other agents move into oversight of the AI, checking its answers and stepping in when it fails. The work is less about speed and more about resolving the things a machine cannot.
The skills the new role needs
The old hiring profile rewarded speed and the ability to follow a script. The redesigned job rewards different things. Problem-solving, because the easy contacts are gone. Emotional steadiness, because the contacts that reach a person are more often the upset ones. Comfort working alongside software, because the agent is now supervising or correcting an AI rather than just using a tool.
This is why 84% of leaders are revising hiring criteria. A candidate who would have been screened out for slower handling times may be exactly right for a role that is now about judgment. Writing skill matters more too, since agents feeding knowledge bases are producing the content the AI reads.
What it means for hiring, training, and pay
Three practical shifts follow.
Hiring. You are recruiting for judgment and composure, not throughput. Job descriptions, screening questions, and assessment tasks all need to reflect that. Hiring volume may also fall, so each hire matters more.
Training. Onboarding built around scripts and quick handling does not prepare someone for a queue of hard contacts. Training has to cover dealing with complexity, working with the AI, and the judgment calls the AI hands over. Existing agents need retraining, not just new hires.
Pay. A harder job is worth more. If the role now demands skills closer to a specialist than a script-follower, pay bands set for high-volume routine work will look low, and the better agents will notice. Metrics need to move too: average handle time made sense when contacts were uniform, but it penalises an agent for spending the time a complex case actually needs.
Where this leaves the agent role
The contact centre agent job is not ending, but the version built around speed and scripts is. The work that remains is harder and needs more skill, and the centres handling this well are redesigning the role, the training, and the pay around that harder work rather than waiting for the change to force itself. If you want the wider picture, see our notes on whether AI will replace customer service jobs and on the business case for keeping humans in the loop.