Re/insurers that view artificial intelligence (AI) as a capacity creator rather than simply a cost-reduction tool can unlock greater value from the technology, according to Martin Henley, CEO of mea Platform.
In an interview with Reinsurance News ahead of the annual Rendez-Vous de Septembre in Monte Carlo, Henley emphasised that using AI to reduce costs can deliver meaningful savings. However, firms that scope it as a capacity creator can process more submissions without adding headcount, improve turnaround times for brokers, and ultimately write more business.
“Growth follows capacity in a way it does not follow savings,” said Henley. “mea Platform, as one example, has processed $450 billion in GWP since inception. Our view is that with AI, operational drag should be a solved problem.
“For 300 years, insurance has scaled the same way: more volume, more people. The industry now has a digital workforce that executes the repeatable work and applies every control, changing what an operating model can look like rather than how fast the existing one runs. Brokers and underwriters get their day back for risk selection, pricing judgement, and client relationships. That is where underwriting performance and profitability come from, and that is where the next twelve months of value will be created.”
Over the past year, Henley explained that the conversation around AI in insurance has shifted from whether it could carry real insurance work to execution and ROI.
He continued, “The industry has heard versions of this promise before: OCR, BPO, system replacements, and a long line of InsurTechs arriving from outside the industry. Each delivered something real in its own lane, but none changed the shape of the operation. Across implementations at scale, the pattern is consistent: the problem was never the individual task. It was the fragmentation between tasks: the inboxes, the queues, the handoffs, the rekeying. Most of the processing time in insurance is waiting time, not thinking time.”
Henley highlighted that what is different now is agentic AI, which, when built specifically for insurance, can own whole processes rather than individual steps within them.
He explained, “Generic tools do not know the nomenclature, the document types, or the exceptions, and so they end up applied narrowly or layered on top.
“Conversely, insurance-native systems can carry a submission from arrival through to a bound policy, or a claim from first notice through to settlement, with exceptions routed to the people who should handle them. Where this is running in production, the gains are showing up in operating cost, in underwriting capacity, and in the time it takes to move work through the operation. That is why we are seeing so much expansion: organisations start in one line of business and extend into the next. The results are proven, but the capability to operate at scale depends entirely on the technology being insurance-native. That is where most programmes stall: a pilot runs in a controlled environment on a clean set of documents, while live operations are far less predictable.”
According to Henley, insurance-native AI can remove much of the implementation burden traditionally associated with new insurance technology.
“The obstacle most firms expect is implementation, and for a long time they were right: for decades implementation would become the innovation itself. The integration programme took over the project, and the outcome it was meant to deliver became overshadowed by simply getting the system connected. Insurance-native AI takes that off the table: no multi-year transformation and no lengthy configuration cycle. What decides the return now is how firms scope the work, and how they measure it,” he said.
