Guidewire's Mullen: Core Systems Must Anchor Agentic AI in Insurance
John Mullen, president of Guidewire, argues that as insurers deploy AI agents to automate decisions and actions, the reliability of their underlying systems becomes more critical, not less. The insurance industry is moving faster than it has in previous technology cycles, with some carriers already scaling AI beyond proof-of-concept. Others are building AI assumptions into budgets to pull organizational change forward.
Mullen sees AI as a general-purpose technology that differs fundamentally from earlier shifts. Previous waves helped organizations scale operations or improve prediction. AI does both simultaneously-it increases scale and expertise in equal measure.
The Business-Technology Divide Collapses
That dual effect will reshape how insurers organize work, make decisions, and connect strategy to execution. Mullen said the traditional separation between insurance professionals and IT professionals no longer holds.
"There is no longer an opportunity where somebody can say, 'I'm in the insurance industry, I'm an IT professional,' or, 'I'm in the insurance industry, I'm a business professional,'" Mullen said. "Those lines can no longer be drawn. You have to be both if you want to navigate and lead in this industry."
The speed of change varies. Some carriers are already past experimentation and into early scaling. Others remain in proof-of-concept mode. Still others are building AI assumptions into budgets to reshape their organizations faster.
But speed alone is not the answer. Mullen repeatedly stressed the need to balance agility with durability. Insurers need velocity in product development, risk selection, pricing, and customer interaction, but they must also preserve the reliability, auditability, and regulatory defensibility required to operate at scale.
Context Is Everything for AI Agents
The deeper question for core systems emerges when AI begins not only to summarize or assist, but to act. Agentic AI agents will require the same business context that human workers have historically drawn from core systems.
"In order to unlock the full power of AI and build an agentic insurance carrier, you have to be running on a modern core platform," Mullen said. "Just like the human actors interacted with a core operating system, the agentic actors will need that same foundational element for predictability, scalability and reliability."
That context includes policy, claim, billing, and underwriting information that enables decisions to be made correctly and repeatedly. It also includes the rules and controls that make those decisions auditable and suitable for a regulated industry.
AI's probabilistic capabilities must be balanced by deterministic insurance rules and durable operating processes. The goal is not merely to generate answers, but to generate answers and actions that are appropriate, repeatable, and improvable.
"Context is everything," Mullen said. "In order to become an agentic carrier, you need to be operating on a modern, reliable core system because the context upon which your agents operate, the context upon which your insurance rules exist, are absolutely fundamental."
A Staged Progression From Answers to Actions
Guidewire's approach operates across several layers. AI agents can first provide answers in the context of a user's work. Then they can provide suggestions, flagging issues or recommending next steps. Finally, they can be configured to take actions inside the workflow.
The company's acquired ProNavigator technology delivers agentic capabilities into Guidewire applications such as ClaimCenter, PolicyCenter, and BillingCenter. These capabilities pull together carrier, customer, claim, policy, and billing context to help workers answer questions like "What should I do next?"
From there, the technology can suggest actions or warnings-confirming special handling instructions, checking policy coverage, identifying potential work injuries, or highlighting changes to claim files. The final stage is taking those actions automatically.
Mullen emphasized that not every process becomes fully autonomous at once. Instead, workflow becomes less visible to the user over time. "Over time, the workflow will move more to the background, and the work and the decisions themselves will move more to the foreground," he said.
Claims and Underwriting as Prime Use Cases
First notice of loss is among the most visible early use cases for AI in insurance. Commercial underwriting is likely to be profoundly changed, especially because of the volume of unstructured information exchanged between brokers and carriers.
"The exchange of underwriting information in the world of commercial insurance between broker and carrier will be the most fundamentally changed, just because of the amount of unstructured data," Mullen said.
Risk selection, pricing, product design, and the exchange of information between primary carriers and reinsurers are other areas where AI can create meaningful gains. Getting data to the right place in the enterprise without the constraints of human conversation will be powerful.
The Overlooked Cost: Adjuster Burnout
Mullen raised a less obvious consequence of AI-enabled claims work. If AI removes administrative burden and reduces the process time between customer interactions, it may also remove the natural pauses that claims professionals have historically had between emotionally intense conversations.
In claims, AI should allow people to focus on work that depends on their expertise rather than merely their presence in the process. That includes delivering a more compassionate and brand-aware claims experience. But it also creates risk.
The traditional claims workflow sometimes created a "natural mental break" for the adjuster after an empathetic exchange with a customer. As AI compresses administrative work, those breaks may become shorter.
"How do you build in the mental health awareness and all of the components that make sure that the claims adjusters who don't have that time between those empathetic moments are cared for?" Mullen said. "It's going to be exhausting, and we have to really think and care for that."
Faster work is not automatically easier work. If AI removes low-value tasks, insurers may also need to redesign jobs so that the remaining human work is sustainable.
The Larger Opportunity
For Mullen, the long-term opportunity extends beyond workflow efficiency. AI can help insurers become more relevant, dynamic, and responsive to customers while preserving the discipline required to underwrite, price, service, and indemnify risk.
The insurance industry has long suffered from a gap between the value it provides and the trust it receives from society. AI will not close that gap by itself, but Mullen believes it gives insurers new capabilities to narrow it.
"For a long time, this industry has suffered between this kind of trust and value gap," Mullen said. "The value that the industry offers to society and the trust that the industry has garnered with society has been a gap too big for too long."
After three decades in insurance core systems, Mullen said he is more optimistic than ever about the industry's ability to improve that relationship. The foundation for change is already being laid.
Your membership also unlocks:
