The insurance industry’s AI conversation is moving past experimentation and toward a more demanding question: where is the return?
Speaking during the “AI Revolution – From Hype to ROI” panel at InsuranceFest 2026, executives from insurers, brokerages and technology providers said the most credible near-term gains are still being found in document processing, workflow automation and employee productivity.
But the more ambitious opportunity may lie beyond cost reduction. Panelists argued that better data, faster underwriting and lower-friction distribution could eventually help insurers write more business, improve profitability and address parts of the protection gap.
Doug Alexander (pictured left), senior vice president and chief technology officer at Upland Specialty Insurance, said the carrier has seen measurable benefits from document extraction.
“What we’re seeing in the AI space is we’re getting a lot of benefit out of document extraction and being able to automate some of that manual, tedious work, and we can measure the ROI with it,” he said.
Upland is also using AI to improve individual workflows and beginning to connect those tools into more agentic processes. However, Alexander stressed that human oversight remains central.
“It’s our position at Upland that we want to make sure that our human expertise is augmented with AI, but not replace it certainly,” he said.
For insurers trying to justify investment, Alexander said narrowly defined operational savings can offer a more defensible starting point than broad claims about future underwriting performance.
“It’s hard to lean in with an AI tool or any project to say ‘We think this is going to improve the loss ratio or the combined ratio’,” he said. “But getting real specific on OpEx has been our approach in the AI tools to start.”
Doug McElhaney (pictured right), chief strategy officer at Applied Systems, similarly described productivity as the “first order of value creation.” He said firms should measure whether AI is genuinely removing time from processes and allowing employees to redirect that capacity elsewhere.
Applied Systems is developing what McElhaney called AI-powered workflows of the future, with the aim of eliminating much of the manual work currently embedded in insurance processes.
“We envision these workflows eliminating 80 to 90% of the manual work that’s associated with them,” he said. “Now that’s not going to happen overnight.”
However, McElhaney said the industry may have only a limited window in which productivity gains remain the central story.
“I think there is a 12-to-18-month window where we can continue to see real gains and real value created just on that first element of productivity,” he said.
After that, the focus could shift toward whether AI can create new premium opportunities and improve the way risks move between insurers, brokers and customers.
Christina Lucas (pictured centre-left), Google Cloud’s global market leader for insurance, said discussions with insurers are increasingly moving from expense reduction toward growth.
“You can only cut expenses so much,” Lucas said. “But you can have unlimited revenue growth.”
She pointed to distribution, retention, customer acquisition and underwriting as areas where AI could increase revenue per employee and improve profit margins.
“Your loss ratio improvements and your underwriting profitability are, I think, one of the big areas that we’ll see growth in the next 12 to 24 months,” she said.
Monica Sanjinez (pictured centre), executive partner and commercial lines leader for Southern California at USI, said the benefits are already visible on the brokerage side through faster client preparation, submissions, proposals and policy checking.
For producers, she said AI can act as a “critical thinking partner,” allowing teams to prepare more quickly and focus on stronger client solutions.
Yet the technology’s success will depend heavily on whether employees trust and adopt it.
“The AI is not replacing people, it’s replacing tasks,” Sanjinez said. “It’s empowering us to do more – faster.”
That distinction may be crucial as firms confront resistance from employees concerned about automation. Lucas said organizations achieving stronger returns tend to combine broad employee access to AI tools with a small number of CEO-led transformation priorities.
Lucas said the approach comes down to two things: giving employees broad, sanctioned access to AI tools like Gemini, ChatGPT, and Claude, and having leadership – starting with the CEO – identify two or three top transformational priorities for the organization to pursue.
Christopher Frankland (pictured centre-right), founder of InsurTech360, said insurers must avoid automating flawed processes simply because the technology is available. The starting point should be workflows with the greatest friction, scale and repeatability.
“Identify the workflows that have the most friction, which ones can we scale, which ones are repeatable and then kind of start with that,” he said.
Longer term, McElhaney believes agentic AI could go beyond assisting employees and begin running entire workflows.
“I can imagine a world where at one point a risk is negotiated by two AI agents because they understand the institution on either side, they understand the rules that they need to follow which you can instrument, and they understand an outcome that both parties are trying to get to,” he said.
