When most of us think of AI, generative tools that answer prompts come to mind. In insurance, underwriting is the process of assessing a client’s risk and deciding whether to offer cover, on what terms, and at what price. In underwriting, a newer class of “agentic” AI is emerging that is less about producing text and more about carrying out decisions and tasks.
Generative AI typically responds to a single, one‑off request (for example, “draft an email to this broker”), while agentic assistants can break a broader goal into steps and then plan and execute a sequence of actions. The key difference is that Agentic AI can operate with far less human intervention, taking on higher functions such as reasoning, adapting and decision making.
An agentic underwriting assistant might read a broker submission, classify documents and extract key data fields before running pricing models. It can then draft quote documentation and broker correspondence, flag portfolio limits and route exceptions to a human underwriter for approval.
Implications for Businesses and Consumers
One article boasts that Agentic AI may cut quote preparation times by 60-99%. Another source reports that reduced times are also matched with high accuracy with AI systems achieving 92-94% accuracy. Research by Mckinsey echoes this optimism reporting that insurers that have adopted AI experience improve their loss ratio by 3-5 percentage points. For consumers, Aviva has indicated that AI in claims processing has reduced customer complaints by 65%. It is important to note however, that results may vary by line and size of business. A skilled workforce, capable of managing advanced technology, is needed for successful implementation.
Two contrasting views
Supporters argue that the goal is augmentation, not replacement. Hyperexponential describes agentic AI as “autonomous, goal‑driven systems” that coordinate tasks while underwriters retain judgment on complex risks and final pricing decisions. This view emphasises that AI models are “collaborators” rather than competitors. This reflects a wider industry view that AI should free professionals from repetitive work.
Not everyone is entirely convinced. Suzzane Bray, Head of Talent and Growth at Convex, raises concerns about identifying clear pathways for junior underwriters, she emphasises redesigning the underwriter role not replacing it. She says “the future of underwriting will depend not just on the talent we attract, but on how intentionally we develop it. Without clear pathways, mentorship and knowledge transfer, we risk hollowing out the pipeline – particularly for women and emerging leaders. This is the moment to redesign underwriting careers, not simply replace roles.”
Other challenges include the risk of bias, opaque decision paths and skills erosion. For commercial brokers, the immediate impact is likely to be more consistent turnaround times.
To conclude, many emphasise the role of governance but as the capabilities of AI models are changing rapidly firms will need to revisit and update their controls regularly.
