Commercial insurance is moving into a period where workforce demographics and technology adoption are changing at the same time. A significant proportion of experienced underwriters and brokers are approaching retirement, particularly in specialty lines and relationship-led segments of the market, and that change is happening just as AI is being embedded more deeply into underwriting processes.
Estimates suggest that close to
They have seen how loss trends develop over time, how wording adjustments alter exposure and how pricing discipline can erode under competitive pressure. That experience is rarely captured in a formal knowledge base - it develops through repeated exposure to complex placements, challenging renewal cycles and difficult claims years that unfold over long periods. It is also embedded within market relationships — a nuanced understanding of how a broker operates, how a client's needs and risk appetite evolve over time, and the extent to which flexibility can be applied responsibly to support sound commercial outcomes. When succession planning is informal or inconsistently managed, a significant portion of that market insight and relationship capital can be lost with the individual.
At the same time, insurers are increasing their use of AI across submission intake, triage, and risk assessment. With growing submission volumes and rising service expectations, a degree of automation is no longer optional — it is operationally necessary. The real consideration for the market is not whether AI should play a role, but how it is deployed, governed, and integrated into underwriting in a way that supports judgement rather than replaces it.
AI should reinforce underwriting capability rather than narrow it
AI delivers meaningful value in how information is captured, structured, and analysed across the underwriting process. It can extract and standardise data from submissions received in multiple formats, enrich files with relevant external data sources, and present a more complete view of the risk at the outset. It also flags missing information, identifies inconsistencies, and triages cases against defined underwriting criteria.
Yes, there is incremental efficiency, but more importantly, AI strengthens consistency, improves data quality, accelerates decision-making, and allows underwriting teams to focus their expertise where it adds the greatest commercial and technical value.
AI cannot replace the seasoned judgement of an experienced underwriter. It lacks insight into how a particular class has responded to regulatory changes, how loss development has impacted past portfolios, and how the nuances of broker relationships should shape negotiation and decision-making. It does not assume responsibility for capital allocation decisions or for determining how much aggregate exposure is appropriate within a given segment. Those responsibilities sit firmly with accountable individuals.
There is a risk that an overemphasis on AI could inadvertently limit the role of the underwriter. If junior underwriters are used mainly to review system-generated outputs rather than to exercise independent judgement, analyze risks, and defend their decisions, their professional development is constrained. Over time, this narrows the depth of expertise within the firm and, ultimately, the market, reducing the capacity for nuanced risk assessment and strategic decision-making.
For AI to deliver its full value, it should support underwriters rather than replace their decision-making. New underwriters benefit from seeing more of the picture, not less. By providing easy access to prior loss experience, portfolio positioning, and comparable risks, AI gives them richer context, while they remain responsible for making the call and articulating their reasoning. This combination of insight and accountability is what truly develops strong underwriting judgement.
Technology must be implemented thoughtfully. Simply layering new tools onto existing processes adds steps rather than reducing them. Insurers that streamline submission handling first find automation more effective, as it enhances a clean process rather than patching a messy one.
As senior practitioners retire, firms face a pivotal choice: use AI merely to handle higher volumes, or leverage it strategically to develop the next generation of underwriters. Automation can improve speed and consistency, but it cannot transfer experience. Firms that treat AI as a tool to enhance and guide underwriting judgement, rather than replace it, can accelerate learning, preserve institutional knowledge, and sustain strong long-term performance of the firm.
I believe strongly that AI is essential to creating a smarter, more engaging workplace - attracting top talent and building the next generation of skilled underwriters.
