The transition to AI underwriting is underway — here’s how insurance companies can prepare

May 12, 2025 | min read

Andreza Pires

Vast quantities of unstructured data cross an insurance underwriter’s desk each and every day. The job of parsing through this data to make the best decisions about risk, pricing, and uncertainty is getting more and more complex. So it should come as no surprise that insurance companies are encouraging their underwriting teams to experiment with generative AI, both by building in-house models as well as using third-party tools.

The global insurance analytics market will reach a value of $29.4 billion by 2032, and the widespread adoption of AI in underwriting and pricing will certainly be a large driver of this growth. Slow underwriting is a huge liability for insurance firms, as it’s a leading reason why SMEs switch their providers. The use of AI can reduce underwriting time by 50%, according to BizTech Magazine’s 2025 analysis, while also reducing fraud, as well as boosting accuracy when it comes to pricing and risk assessment.

However, with all the opportunities that the shift towards AI brings, there are also significant risks for firms. With all that’s at stake, it’s worth taking a look at the top considerations and potential challenges that insurers need to know when it comes to incorporating data intelligence into their underwriting and pricing models. The goal for firms should be to stay competitive and compliant, all while maximising the utility AI can offer.

1. Don't get left behind

According to patent analysis, more than 380 companies — from leading insurance companies like State Farm to smaller startups — are integrating AI into their underwriting practices.

Industry leaders are already utilising this technology. AXA XL is a prime example of the growing use of AI in insurance, automatically getting data out of and having a proper look at property site surveys. Their "Digital Risk Engineer" goes a step further, using connected building systems to monitor the state of assets, which helps with underwriting and means they can process over 10,000 reports a year at a much faster rate. On top of that, Insurtech firms like Zego and Oscar show just how powerful this technology is for offering cracking good prices. As a result, the entire insurance sector is feeling the pressure to get on board with AI or risk being left in the dust.

2. Risk is on the rise

Trade wars, geopolitics, climate shocks, regional sanctions, natural disasters and more — all of these events are becoming more common. In fact, 40% of marine underwriters reported increased claims due to geopolitical shocks in 2024. All of these events make it more difficult for insurers to model risks accurately and underline the need for AI models to help keep up with the complexity and unpredictability.

3. Poor quality data leads to poor quality underwriting

A report from Forrester notes that 60% of insurers struggle with fragmented data split across various sources, from IoT devices, customer claims history, and third-party APIs. Other data, such as property prices can be outdated or low in quality. These inadequacies prevent underwriters from getting the full picture to assess risk accurately. In these situations, AI can only help so much — data modernisation projects must be a priority to make full use of AI.

4. Insurers still need a human in the loop for complex decisions

Insurers face a slew of overlapping regulations when it comes to underwriting, from the FCA’s Consumer Duty to the European Union’s AI act. The latter means that insurers have a duty to justify the pricing decisions that AI has made. While AI can help speed things up, fully automating complex decision-making is more likely to lead to non-compliance. There needs to be a balance of integrating AI while still keeping a human in the loop.

5. The skills gap is real

The adoption of AI requires skilled data scientists, but 55% of firms say they lack this relevant expertise, according to Deloitte. In addition, with AI in the mix, junior underwriters need more mentorship and guidance to know when to step in and when to allow models to make decisions.

The great transition to AI underwriting is underway, and insurance firms don’t need to wait until they can build in-house solutions in order to get started. CI&T’s expertise in AI, data integration, and legacy modernisation directly addresses many of the challenges that insurance underwriters are facing. Our suite of tools and services that can help make this transition smoother, without having to build in-house or proprietary tools.

Our AI-driven underwriting solutions help firms consolidate fragmented data silos into unified platforms. Modernising legacy data systems means you benefit from AI’s full potential by giving it up-to-date and accurate claims data from which to make decisions. In addition, it boosts compliance and makes it clearer when a human being needs to get involved.