Videos
How Credendo Is Applying Agentic AI to Credit Insurance Decisions
Insights
- In credit insurance, AI value comes less from conversational interfaces and more from scaling faster, more consistent, and explainable risk decisions.
- The shift from predictive and generative AI to agentic AI marks a move from productivity support to partial ownership of end-to-end workflows.
- Human expertise becomes more valuable as AI absorbs data extraction, orchestration, and real-time monitoring.
In this conversation, Jan-Pieter Laleman, Chief Data Officer at Credendo, and Jon Holvoet, Chief Technology Officer, explain how AI adoption in trade credit insurance is moving beyond experimentation. They describe why mature use cases focus on lead detection, risk profiling, and decision automation rather than customer-facing chatbots, and how Credendo is progressing from predictive and generative AI toward agentic systems that orchestrate tasks across underwriting, claims, and risk management. Drawing on their collaboration with Infosys, they discuss how explainable models, autonomous agents, and cloud-scale platforms are enabling faster, more consistent credit decisions, while keeping human accountability firmly in place.
Jan-Pieter Laleman:
The current state of AI in the credit insurance space is in the end similar to the path of AI development in the insurance sector in general, but of course with specificities for our lines of business.
So thereby AI in credit insurance is less about, for example, flashy chatbots because we are a very customer focused company, but more about lead detection and in general better risk decisions that scale.
So the mature players are using AI to improve lead detection, depth of risk profiling, portfolio management, decision automation and technology wise we also see a mix.
It's not about large language models alone, different flavors that exist, but it's about combining machine learning with explainable decisions are important with the current boom in agentic AI for specific task handling, for example.
And this is also where we are collaborating today with Infosys, where we are working on different use cases.
And for example is debter risk. We are using today AI to help underwriters combine all the internally available data together with all the externally available data that you can find. So combine painless experience that you have yourself with external data that you have to really turn that into sharper structured overviews and there the essential part is really speed efficiency on the one hand but on the other hand consistency and accuracy.
Jon Holvoet:
AI and trade credit insurance gives us great additional benefits. On the one hand, with regards to unstructured data, it allows our underwriters to focus more on the tasks that matter instead of putting a lot of manual time and effort in extracting data from sources that are not automated.
On the other hand, by enabling all of our employees with their own personal AI assistant, we help them tremendously in speeding up certain tasks that are very mundane and time consuming with the assistant that has access to the full history and context of the situation.
Jan-Pieter Laleman:
So the way that AI insurance is evolving is we're going through multiple phases. So phase one was the mode of predictive AI. So we're using AI machine learning for a long time for scoring, segmentation, fraud flags. So they have a very powerful usage, but at same time it's very narrow in use. Then we went to phase two. We have all the different large language models that came out and we start using them for drafting emails, summarizing files, and that were elements that were really supporting your underwriters, supporting your claim handlers in helping with productivity. But now we're really in the third phase, the phase of agentic AI. So it's no longer about generating text. It's now really about planning and executing specific tasks across tools.
So for example, it can be retrieving data, running a check, proposing a decision, opening a case for a person to review and escalating an exception in a certain way. And there the difference is really ownership on the workflow almost. And that is really possible because the platform part is becoming much more mature.
So you have the major cloud providers together with companies like Infosys that are making the building agents, deploying agents more production grade and are offering better toolkits around it. And that is really what insurers need today.
Jon Holvoet:
The use cases we are exploring for AI agents in trade credit insurance turn around autonomous orchestration all across the value chain of our company, from underwriting to claim handling up until risk management. By having these autonomous agents focus on tasks like, for example, extracting missing debtor information, dynamically adapting credit limits, or giving insights and early watch alerts for real-time portfolio monitoring, we can make sure that our people can focus on the tasks that matter the most and that benefit our clients the best.
