Berlin, April 16, 2026. Four people are on stage at the Customer Focus Summit #6: Isabell Neubert, who is responsible for AI IT governance at coeo, along with three colleagues from the contact center. Gelsomina Ceruti, Anna-Lena Jungbluth, and David Winkelmann work with cAI—coeo’s AI assistant—every day. Their talk isn’t a glossy sales pitch. It’s an honest assessment of their day-to-day work: What works, what doesn’t, and where is this journey headed?

The question of whether AI will replace humans in debt collection is a major concern for the industry. At coeo, they’ve stopped discussing it in theory. They’ve simply gotten started.

From Phone Call Chaos to Focused Work

Before cAI came along, the contact center was a different place. David Winkelmann, who was there from the very beginning, recalls: “Back then, the volume of phone calls was significantly higher. Simple inquiries that came up day after day: ‘How much do I owe?’ ‘Can I set up a payment plan?’ ‘Can I get a confirmation letter?’ Today, cAI handles the majority of these inquiries before they even reach an agent.”

“The discussions are now focusing more on more specialized cases—either handling parallel cases or appeals that go relatively deep. We then take on those cases.” — David Winkelmann, coeo Contact Center

For Anna-Lena Jungbluth, this means one thing above all else: more time for what really matters. No longer sending out a reminder letter every two minutes, but instead corresponding with attorneys and debt counselors, delving deeply into cases, and solving real problems.

“I’d say this makes my job very enjoyable. cAI takes care of so much that I can focus on the things that really matter.” – Anna-Lena Jungbluth, Contact Center coeo

The figures confirm what we see in our day-to-day work: Over the past four weeks, cAI has processed 64.04 percent of all incoming emails using “Text-Flow,” its AI-automated inbox processing system. The impact is also evident in concrete figures for “Voice Agent,” the AI-based voice assistant: The average validation rate was 66.2 percent, and the forwarding rate was 43 percent.

cAI detects, documents, and takes action

cAI is more than just a telephone assistant. It verifies the caller’s identity, documents the caller’s request in the notes field, automatically sends out correspondence, rejects invalid installment offers, and updates the file with information. For Gelsomina Ceruti, this was one of her first real “aha” moments:

“My first ‘aha’ moment was when I identified the issue using cAI. And then, looking at the timeline, I realized, ‘Oh, right, cAI had already flagged that. I can mark it as completed and move right on to the next case.’” – Gelsomina Ceruti, coeo Contact Center

cAI also plays a role in case management: It handles routine decisions regarding emails, checks requests for admissibility, and provides case managers with a clear overview of what has already been completed. This not only saves time; it changes the way the team thinks and sets priorities.

And the debtors? They’re responding surprisingly well.

One concern often arises with many AI implementations in customer-facing roles: How will people react when they realize they’re talking to a machine? At coeo’s contact center, the picture is nuanced but largely positive.

Many debtors feel relieved when simple tasks have already been taken care of in advance, according to Anna-Lena Jungbluth. A letter confirming the matter has been sent, the data has already been verified, and the amount owed has been communicated. When they then reach a staff member, the conversation doesn’t start from scratch.

And what about those who are annoyed because they didn’t want to talk to an AI? The team is familiar with that, too. But even those conversations often turn out differently than expected.

“Because they’re annoyed by the AI, they’re more likely to strike up a conversation: ‘Oh, nice to have a real person on the line.’ And then they’re actually even more relaxed, because it’s such a relief.”
– Gelsomina Ceruti, coeo Contact Center

David Winkelmann puts it even more bluntly: cAI filters out the emotional aspect. Anyone who comes in with a real problem is directed to an employee who has the capacity to truly address it.

Governance: The Underestimated Challenge

What becomes particularly clear on the stage at CFS #6 is that the technical implementation of cAI is only part of the story. Isabell Neubert, who heads the AI IT Governance division at coeo, identifies the real complexity:

“Governance now means bringing knowledge down to the level of every employee. I need to raise awareness. And I also need to ensure that approved tools are used within the permitted parameters.” – Isabell Neubert, AI IT Governance coeo

AI is one of the biggest topics right now: Existing IT systems are now integrating their own AI modules, and employees prefer to use their favorite AI system. In the context of the EU AI Act, this gives rise to specific liability risks, especially if AI systems are used in high-risk areas such as HR or credit decisions.

At coeo, the decision was made not to sidestep these issues but to address them in a structured manner: risk assessment, clear rules of use, and employee training. The process isn’t over yet, says Neubert, but the direction is clear.

What’s next?

At the summit, the question was also raised as to what cAI should learn next. David Winkelmann’s answer is surprising: He’s thinking not only about the debtor side, but above all about internal professional development:

“Internally, there’s definitely room for improvement there. For example, when it comes to training and developing new employees, we can use various case studies generated by AI from the data.” – David Winkelmann, Contact Center coeo

For Gelsomina Ceruti, the most important thing is something else: Those who understand cAI are less hesitant to engage. Not just in the contact center, but externally as well. This understanding will become even more important with the next phase of expansion: the Decision Navigator, a new core component that cAI has been developing since the summit.

It is designed to determine the so-called “Next Best Action”—that is, the most appropriate next step in a specific debt collection case. Traditional scoring models provide only a summary metric for this purpose. The Decision Navigator, on the other hand, is designed to take the timeline of a case into account: Whether someone has paid a large sum once or has made small, timely payments over several months makes a difference for the forecast, even if the total amount is the same.

The logic was initially developed for MB scoring—that is, determining which cases should be escalated to a payment order—following approximately nine months of foundational work on the database. Going forward, the Decision Navigator is intended to serve as the central decision-making logic for the entire cAI ecosystem: the Voice Agent, Text Agent, and the response expert for case handlers will be able to access it. At the same time, its insights feed directly into Ikaros’s workflows.

Proof, not a promise

What makes the talk at Customer Focus Summit #6 so special: It’s not a presentation of concepts. It features three people who work with cAI every day and speak their minds. Not because they have to, but because they’ve experienced it firsthand.

cAI hasn’t revolutionized coeo’s contact center. It has quietly transformed it—step by step, feature by feature, call by call. And it’s not done yet.

To the Panel Discussion

Isabell Neubert, Gelsomina Ceruti, Anna-Lena Jungbluth, and David Winkelmann discussed AI governance, compliance, and day-to-day operations in contact centers at Customer Focus Summit #6 on April 16, 2026, in Berlin. The full talk is available in the CFN media library.

Cover Image © Pete