Hundreds of decisions are made every day. Should this case go to a payment order or not? What’s the next logical step? Traditional scoring models provide an answer, but often the wrong one, because they don’t know the story behind the number.

cAI Technology has been working on a different solution for just under a year. The Decision Navigator is intended to be the result.

One Industry, One Unpaid Bill

This idea has been a hotly debated topic in the debt collection industry for years. In a complex debt collection case, how can one automatically and reliably determine what to do next? Not based on blanket rules, but on the specific case, its history, and its probabilities.

“The Next Best Action—that is, determining what the most effective next step is in this specific case—is something like the Holy Grail that everyone would love to have at their disposal, but which hasn’t been easy to achieve so far.”
– Stephanie Dittrich, Head of AI Development at cAI Technology.

Standard solutions on the market provide general scores. They aggregate data, condense it into a single number, and use that to provide an assessment. What they cannot do is understand the timeline of a case, identify patterns in payment behavior, and derive an individualized recommendation from that information. That is exactly what the Decision Navigator is designed to do.

The difference lies in the passage of time

A classic scoring model considers aggregated data points: a single number that summarizes a situation. The Decision Navigator is designed to think differently. It will analyze sequential information in chronological order.

Stephanie Dittrich illustrates this with a simple example: “There’s a difference between someone paying 100 euros once a year and someone paying 10 euros on time every month for the past ten months. The total amount is the same, but the likelihood of another payment is completely different.”

The system is designed to draw on hundreds of data points when making a decision prior to filing a petition for a default judgment. In addition, state-of-the-art AI methods are used, including approaches inspired by quantum research, to better map complex decision spaces. Humans are intended to remain in the loop: Where decisions have consequences for people, humans set the cutoff. The AI provides the probability, while the basis for the decision remains transparent and traceable.

Nine months of basic research

The development began with a specific use case: MB scoring—that is, determining which cases should reasonably be escalated to a payment order. What sounds like a narrowly defined problem required extensive preparatory work: nine months, many explored approaches, and intensive work on the data foundation.

That was a deliberate choice. Anyone who wants to build a universal decision-making logic needs a solid foundation. cAI took the time to truly understand the entire database before moving on to the next steps. The Decision Navigator is therefore not a hastily developed automation tool, but rather the result of structured basic research.

The go-live is imminent. The expectations are clear: When it comes to decisions on payment orders, the Decision Navigator is expected to identify non-payers with significantly greater accuracy than the previous rule-based scoring system. The goal is no longer greater automation, but rather fewer incorrect decisions and, consequently, lower unnecessary litigation costs.

Debt Collection That Puts People First

The Decision Navigator is designed to be not only more efficient, but also fairer. Ideally, the debt collection experience will be more pleasant for the debtor because the measures are tailored to the individual rather than being standardized. “In the best-case scenario, we support the debtor exactly where they are and help them reduce their debt,” says Dittrich.

This is an effect that should also pay off for clients: A fair process increases the chance of winning the customer back once the proceedings are concluded. When it comes to “Next Best Action,” the focus therefore goes beyond the mere collection rate. It’s about determining which action will have the greatest impact at the right time through the right channel—for everyone involved.

A central nervous system for the cAI ecosystem

The Decision Navigator is not intended to be a standalone module. It is designed to serve as the central nervous system for the entire cAI ecosystem. Other modules should be able to access its decision-making capabilities: the Voice Agent for phone calls, the Text Agent for emails, and the Response Expert for case handlers. All of them should be able to flexibly adapt their responses based on the decision criteria provided by the Decision Navigator.

Yet it is designed to be entirely independent. The Decision Navigator is intended to be able to integrate its insights directly into the workflows of Ikaros, coeo’s proprietary process system, independently of the rest of the ecosystem. Decision logic and operational implementation are thus directly intertwined.

What’s Next

The ultimate goal is a universal decision-making unit—a module that helps users choose the most effective course of action at any time and in any situation throughout the case process, not based on standardized workflows, but on the specific case and the person behind it.

“In two years, I envision a module that can help us choose the most effective course of action at any time and in any situation throughout the case process, so that we don’t have to resort to standardized workflows that cannot take into account the unique nature of the specific case and the person behind it,” says Stephanie Dittrich.

“That would truly be a revolution in receivables management.”

Cover image © Garun Studios