When legacy systems slow down AI: new options for insurers

Artificial intelligence (AI) has been in the insurance industry for some time already. Insurers are investing heavily in AI and are developing new applications and pilot projects, with 82% already regard artificial intelligence as a strategic priority. At the same time, according to a study by Roots, only around 7% of AI initiatives make it past the pilot phase.

Why is that?

When we talk to customers about their difficulties, the situation becomes clearer. It’s not about the vision, nor is it about the strategy – implementation often fails due to the systems used.

Most of these insurers work with software systems designed for all kinds of challenges. It sounds good and offers some advantages, but with AI, such a big system becomes a prison for good ideas that can be implemented with AI. 

The reason is that any adjustment to the system requires a project that sets all areas of the insurer in motion. New data structures, new interfaces, new logic and the AI advantage of training the system to successfully complete the tasks. Everything has to be squeezed into the software. But it wasn’t designed for that. Furthermore, it also struggles to deal with interfaces and open architectures. 

And the tricky thing about this situation is that the current system fulfills an important prerequisite of IT – it works reliably. The more reliable the current system is, the less pressure there is from IT to change it. Nobody wants to touch the tried-and-true system. Business pressure results in the testing of POC after POC, and yet it is still stuck between legacy stability and AI ambition.

What we see in our customers
Insurers who work with us do things differently. They rely on focused, modern systems. Not “all-in-one,” but specialized. Each feature optimally suited to its purpose.

The results are impressive: four of the five most efficient insurers in Switzerland are our long-standing clients (BAG study 2024). 

You don’t need any huge replacement projects. AI is integrated where it really makes sense – not as a retrospective add-on, but as an integral part of the architecture.

You get clean data structures, open interfaces and real learning skills in a clear, controlled environment. Not only is it technically more elegant. It is much more economically efficient.

The insight behind it
Complex systems are often so strongly integrated that it creates a problem. Any change requires a project, and the lack of awareness of interfaces significantly weakens the use of AI.

Focus is the answer, and it pays off directly: faster invoice verification, better data quality, measurable cost savings and more flexible adaptation to regulatory changes.

That’s exactly where we’re investing at BBT
We build software for clarity and stay focused on our core competencies. This enables us to create clear, traceable and measurable added value for our customers. 

We see that the market is re-establishing itself. Insurers who work with modern, focused systems get a head start. They’re more agile. More flexible. More profitable. The others, those sitting on legacy and hoping, will lose out.