CX Leaders Reinvest AI Gains in Coaching to Build the Next Generation of Contact Center Talent

Contact center agents have always built judgment through repetitive tasks, and AI now handles many of them. That leaves leaders looking for another way to show new hires what a good answer looks like. Drafting a detailed customer response or building out a knowledge base used to take hours or days. AI can do it in minutes, which saves time but also removes the practice that taught agents to tell a strong answer from a weak one.

Roman Bodammer is Contact Center Lead at one of the largest European commercial and residential real estate companies, where he oversees tenant service and an internal IT service desk. Over more than 20 years in customer operations, he has worked with organizations including PayPal, Deliveroo, Foundever, and Transcom, and spent nearly a decade as an interim leader and consultant. In 2011, he began building a work-anywhere model for agents, years before the pandemic pushed the industry there.

“If we don’t keep investing in the skills that help us distinguish between okay and good, those skills will be lost over time,” Bodammer says. For managers, that means setting aside dedicated time and budget for training.

Training off the clock

Bodammer sees the same pressure in professional services. AI takes on a growing share of the research junior consultants and lawyers once did, and that research was how they learned the job. Some firms are now carving out separate time to train their early-career staff.

Contact centers face it too. AI resolves many simple questions on its own, so new hires get fewer easy calls to practice on. Without regular feedback, they can feel confident on standard cases but unsure in the gray areas, where one call can decide whether a customer stays. Quality programs can also drift toward checking compliance and away from teaching agents how to think through a situation. “You need to pay more attention and invest even more in quality management and in teaching this agent what good really looks like, because he never learned it on the job,” Bodammer says.

Across the industry, CX leaders are rethinking agent roles, with more of the remaining work focused on complex escalations and knowledge management. Many are also deciding where the hours gained from automation should go.

Frontline co-builders

Setup work that once took months, like documenting processes and governance rules, can now be done in days. One way to spend that time is to involve human agents in building and improving the tools they use every day. “The idea is to use the productivity gain from these tools to keep investing in the improvement of these agents, and to encourage them to co-develop the tools and the use cases. It’s mainly about reviewing together what’s happening, reviewing the results together, and helping the machine learn the better way,” Bodammer says.

Frontline staff are well placed to do this work. They see firsthand when an AI-drafted response misses the customer’s point or when a help article is out of date. When they review those outputs with a supervisor and explain what they’d change, they practice the same skill QA reviews are meant to build. Their corrections also feed back into the system, so the tools get better over time.

Reps can also help choose which tasks the team automates next. They know which requests are simple and which ones need a person to step in. Involving them in those decisions makes the tools a better fit for the work on the floor and shows the people taking calls where their own judgment matters most. Managers can build this into the regular schedule, with set time each week for the team to flag problems and suggest changes.

Measured by results

Bodammer judges agents on what they deliver, a habit that dates back to his first remote teams in Berlin. When he needed Greek and Dutch speakers, he hired them where they already lived. He also opened flexible shifts to parents and people with limited mobility who wanted extra income but couldn’t commute to an office. Today his agents work across Europe and Israel.

Remote work took longer to catch on across the industry, since contact centers run on tight productivity targets and many teams focused on near-term priorities. Managing a distributed team works best with clear outcome targets. Bodammer looks at whether calls get answered on time and whether customers leave satisfied. “I don’t want to pay someone for warming up a seat. I want to pay someone for outcome. If I have someone who’s really good and able to deliver the outcome in half the time, amazing,” he says.

Outcome-based measurement matters even more once AI handles the simpler requests. The calls left for human agents are the harder ones, and how well people handle them shows up directly in service levels and customer satisfaction. That makes preparing people for those calls a bigger priority.

Contact centers that don’t reinvest AI’s time savings in their people may feel the cost years later, when today’s new hires move into senior roles without the instincts their predecessors built. Teams that put part of those gains back into coaching give agents a way to build those instincts on purpose. “Things that you learned on the job yesterday are things that you would need to learn actively tomorrow,” Bodammer says.