Where the Line in (Re)Insurance Falls Between AI and Human Judgment 

83% of insurance leaders say AI should perform the repeatable operational work. 86% say people must retain authority over the consequential decisions. Both findings come from our research with ISG, published this month, and together they delineate the line that the market has drawn on AI in operations. 

Why the Line is Drawn at the Repeatable Work 

Munich Re’s internal studies put underwriters at up to 70% of their time on administrative tasks rather than risk assessment. In different sectors, the items for review are different, but it ultimately amounts to the same result; chasing missing information from a broker, re-keying values from a submission into the policy system, checking whether a schedule is complete, and so on. 

The 83% stat is relatively straightforward once you understand what is meant by “the repeatable.” None of that work demands expertise to be completed, yet all of it consumes the week of the most experienced and talented people in the organization. 

Where Humans Own the Work

The research is crystal clear about where the line is between what AI handles, and what should still be handled by industry talent. Consequential decisions, those that create true company distinction, stay with people: what to price, what to decline, how to structure a coverage call, when to back a broker relationship. 

At mea, we refer to this as an organization’s “secret sauce”; the decisions that drive competitive advantage. Since mea’s inception, our belief has been that the consequential decisions, underwriting talent, and areas of expertise that make insurers unique are where they should be competing. That is where AI can amplify output, but it is not where AI is needed for automation. 

What Happens When the Repeatable Work Moves? 

According to the research, carriers estimate that roughly one in nine broker submissions is declined or left unquoted because operations cannot keep up; on risks they had the appetite to write. 

As mea CEO Martin Henley put it: “The prize on the other side is growth. Carriers are turning away business they had the appetite to write because operations cannot keep up, and winning that back is worth more than any saving.” 

Across mea deployments, organizations see a 40% average increase in underwriting capacity and 75% faster speed to quote, with operating costs down by up to 60% in targeted workflows. An underwriter who gets the week back assesses more risk, quotes while the opportunity is still live, and returns a credible answer to the broker before a competitor does; resulting in more output, larger capacity, and efficiency that can compound across the business. 

The Operational Insight: The Line Between AI and Human Judgment 

Insurance has spent a long time treating operational drag as a fixed cost of doing business. Our research with ISG suggests this is no longer the case; it’s the same reason 96% of respondents said they have an operating model redesign on their agenda. 

When AI is given the repeatable work, the only thing taken away from underwriters is the work that never needed their judgment. That is an outcome worth rebuilding the operating model around. 

At mea, our platform has processed more than $450B in GWP since we launched in 2021, with 30+ clients live today across 20+ countries. We run agentic AI across underwriting, claims, finance, and broking at 92%+ extraction accuracy out of the box, with contractual accuracy guarantees, so you can be certain your people will be focused strictly on the work that their expertise necessitates. 

AI agents own the repeatable. People own the consequential.