Artificial intelligence is reshaping insurance underwriting and claims, but it is strengthening the role of agents and brokers rather than eliminating it, according to Rob Galbraith (pictured), founder and chief executive officer of Forestview Insights, an independent insurance consulting, research and training firm. Galbraith, author of The End of Insurance as We Know It, said the industry has moved through several distinct waves of technological change since the early 2000s and is now entering a period defined by AI sitting at the center of business strategy rather than functioning as a peripheral tool.
“In the past, AI was just another tool in the toolbox,” Galbraith said. “Increasingly, I'm seeing AI front and center, at the heart of organizations, and companies grappling with what that means from a technology, human capital, competitive and value proposition standpoint.”
Galbraith said insurance differs fundamentally from industries like travel and ground transportation, where technology eliminated transactional friction rather than genuine advice. He argued that the stakes of a poor decision in insurance, an infrequent purchase that functions as both a financial and legal contract, are too high for AI to fully replace the counsel that agents and brokers provide.
“If you make the wrong purchase on your own, you can't return a policy after a denied claim,” he said. “It's complex, it's nebulous, and the cost of getting it wrong is high. That's very different from choosing the wrong airline or rideshare company.”
Instead, Galbraith said AI is enabling far greater pricing and underwriting precision. He pointed to telematics replacing broad proxy variables such as age, gender and marital status with direct observation of individual driving behavior, citing his own daughter's experience getting a learner's permit as an example of how granular, AI-driven risk assessment is replacing generalized actuarial assumptions.
On technology strategy, Galbraith said the traditional build-versus-buy debate has become obsolete, replaced by a sharper question of partner versus purpose. He recounted leading a wildfire risk modeling project at a carrier where an in-house team initially outperformed every commercially available model, only for the firm to eventually pivot to outside vendors once dedicated catastrophe modeling companies like AIR and RMS could out-invest and out-data the carrier's internal effort.
“What is core to your competitive advantage, your purpose? Keep that in-house,” he said. “Everything else, partner.” He added that the rise of low-code and no-code platforms is now allowing subject matter experts, rather than dedicated technical teams, to design and maintain new insurance products directly.
Looking toward 2027 and beyond, Galbraith said the pace of technological change in insurance underwriting and claims automation continues accelerating in step-changes rather than gradual progress, an idea he compares to Dick Fosbury's revolutionary high-jump technique at the 1968 Olympics. “That's the mode we're in now,” he said. “ChatGPT comes along and suddenly AI is at a completely different level. Everyone refines it, and then there's the next leap. Those Fosbury Flop moments are happening more and more regularly.”
Galbraith referenced a 10-10 rule, under which emerging technology typically takes roughly a decade to move from concept to practical use, then another decade to scale broadly, suggesting agentic AI may only now be entering its scaling phase. He noted that cloud computing and software-as-a-service tools, covered extensively in this publication's reporting on technology adoption among smaller carriers and brokers, have already leveled the playing field for niche and smaller players that previously could not afford enterprise-grade systems.
Still, Galbraith said true competitive disruption within insurance, the kind that displaces legacy incumbents the way ride-hailing displaced taxis, has yet to materialize at scale, pointing to insurers' ongoing coverage of insurtech challengers and their market performance as a useful gauge of how incremental that disruption has been so far. “We're prone to assuming everything is incremental, everything is linear, and the big keep getting bigger,” he said. “I don't think that'll always be the case.”
