OnClarity Insights
The front line becomes the fixers
As AI takes the routine contacts, the job of the people who stay changes from answering to fixing. Companies that cut agents before they redesign the work end up paying for the same problems twice.
6 min read
The short answer
Redesign the work before you cut the headcount. As AI handles routine contacts, move your best agents into three jobs: fix owner, knowledge owner and AI supervisor. Measure them on fixes shipped, repeat cost removed and time to know, not handle time. Then let the team shrink, if it needs to, through attrition and new roles rather than layoffs first.
Why is cutting agents first the wrong order?
Because the knowledge of where problems come from lives in the people you would cut.
Your longest-serving agents know which billing screen confuses customers, which delivery partner causes the late orders, and which policy line nobody can explain. Very little of that is written down. Cut first and that knowledge leaves with them, while the problems stay. The AI then answers the same questions every week, and the same customers keep coming back.
The industry is starting to see this. Gartner predicts that by 2027, 50% of companies that attributed headcount reduction to AI will rehire staff to do similar work, “but under different job titles” (February 2026). In other words, the work comes back with a new name. Gartner’s analysts have also observed that many organisations are not cutting agents because AI is ready to take over, but to fund the AI.
Most service leaders are already choosing redesign over removal. In a Gartner survey of 321 service and support leaders, 85% said they are expanding agent responsibilities, while 31% had implemented or were planning layoffs in response to AI (press release, April 2026).
Source: Gartner survey of 321 customer service and support leaders, September to October 2025 (press release, 28 Apr 2026).
That does not mean teams are not getting smaller. The same survey found 63% are reducing front-line headcount gradually through attrition. The question is not whether the team changes. It is whether it changes by design or by subtraction.
What new jobs does the front line take on?
When AI takes the routine contacts, three jobs appear that AI cannot do on its own. Each one turns front-line knowledge into problems that stay fixed.
The fix owner
Takes a repeat problem from the queue to its root cause and follows it until it stops. A fix owner does not just tag the contact. They work out why customers keep asking, take the evidence to the team that owns the cause (product, billing, logistics, policy) and check that contacts actually fall after the change. This is the person who closes the loop.
The knowledge owner
Keeps the answers right, for people and for AI. Every AI agent is only as good as the knowledge behind it, and that knowledge goes out of date every time a price, policy or product changes. Gartner reports that 58% of service leaders aim to upskill agents into knowledge management specialists (February 2026). The best candidates are the agents who already know which answers are wrong.
The AI supervisor
Reviews what the AI did and takes what it could not handle. An AI supervisor checks the conversations the AI closed, spots answers that were wrong or incomplete, takes over the cases that need judgement or empathy, and feeds what they find back to the knowledge owner. They are the reason customers can trust the AI, and the safety net when it gets something wrong.
These are not new departments. In most organisations they start as part-time responsibilities for a small group of experienced agents, and grow as AI takes more of the routine volume.
How do you measure the new front line?
Handle time and contacts per hour made sense when the job was answering. They reward short conversations, not fixed problems. Measure the new jobs on what they change:
| Job | Measure it on | Stop measuring it on |
|---|---|---|
| Fix owner | Fixes shipped; repeat cost removed (repeat contacts at baseline minus after the fix, times cost per contact) | Contacts handled |
| Knowledge owner | Wrong or missing answers corrected; time from a policy change to updated answers | Articles written |
| AI supervisor | Time to know for new problems; share of AI-closed conversations that came back | Handle time |
Two rules make these measures fair. Count a fix only when the outcome you wrote down before the change has happened and held. And report repeat cost removed in money, so finance sees what the front line is worth.
What does a 2027 transition plan look like?
A plan a CX leader and an HR leader can start this quarter:
Map the work. List your top 20 contact drivers. Mark which ones AI can answer, which ones need a fix at the source and which ones need a person.
Start with a pilot team. Pick six to ten experienced agents who already explain problems well. Give them one fix owner, one knowledge owner and one AI supervisor brief, part-time, for 90 days.
Train for the new work. Root-cause analysis, writing and maintaining knowledge, reviewing AI conversations, and presenting evidence to other teams.
Build the career path and the pay. The new roles need a level and a salary band, or your best people will treat them as extra work and leave.
Tell agents early and honestly. Explain what is changing, what the new roles are and how people can move into them, before the rumours do it for you.
Shrink by design, not by subtraction. Let attrition and the new roles absorb the volume AI takes. If reductions are still needed, make them after the work is redesigned, not before.
The honest trade-off
Some roles will go. AI does reduce routine volume, and the new jobs need fewer people than the old queue. Not every agent wants to own fixes or supervise AI, and not everyone will suit it. Reskilling costs money and time before it saves any, and the first quarter will look more expensive than a straight cut. The difference is what you keep: the people who know where your problems come from, and a front line that makes those problems stop.
Which of these three jobs could your best agents start next quarter?
Questions people ask
Will AI replace customer service agents?
Not entirely. AI is taking more of the routine contacts, and teams are shrinking, mostly through attrition. But Gartner found 85% of service leaders are expanding agent responsibilities, and predicts half of the companies that cut staff because of AI will rehire for similar work by 2027.
What new roles are emerging in contact centres?
Three are appearing as AI takes routine contacts: fix owners, who take repeat problems to their root cause; knowledge owners, who keep answers accurate for people and AI; and AI supervisors, who review what the AI did and take over what it could not handle.
How do you reskill contact centre agents for AI?
Start with a small pilot team of experienced agents. Train them in root-cause analysis, knowledge writing and reviewing AI conversations. Give the new roles a level and a pay band, measure them on fixes rather than handle time, and expand the programme as AI takes more volume.
Sources
Gartner newsroom, Gartner Survey Finds 85% of Service and Support Leaders are Expanding Human Agent Responsibilities Despite Expectations of Mass AI Layoffs, 28 Apr 2026 (survey of 321 leaders, Sep to Oct 2025).
Gartner newsroom, Gartner Predicts Half of Companies That Cut Customer Service Staff Due to AI Will Rehire by 2027, 2 Feb 2026.
Gartner newsroom, Gartner Predicts Over 50% of Customer Service Organizations Will Double Their Technology Spend by 2028, 31 Mar 2026.
Gartner newsroom, Gartner Survey Finds 91% of Customer Service Leaders Under Pressure to Implement AI in 2026 (58% aim to upskill agents into knowledge management specialists), 18 Feb 2026.
