Guidewire has launched a new platform approach to agentic insurance, giving insurers an open, secure and complete foundation for putting AI agents to work across the insurance lifecycle, from product design and pricing to claims. Announced at its Connections conference in Las Vegas on October 27, 2026, the approach brings together the applications, data, workflows and controls insurers need, and it is presented as a practical model for moving from isolated AI experiments to governed agents that improve the speed and precision of insurance decisions and actions.
Agents Where The Work Already Happens
The central idea is simple: agents should run where insurers already run their core operations, not in a separate AI layer bolted on afterwards. Guidewire agents run on the same trusted platform as its core applications and work with an insurer’s current data and workflows. For agents to work effectively, the company argues, they need to know what is true, what has already been done, and what they are allowed to do, and the platform’s controls enforce those boundaries with consistent security and a clear audit trail that remains current and secure through ongoing updates.
“Insurance is a business of decisions and actions made in context,” said Mike Rosenbaum, chief executive officer of Guidewire. “By embedding agentic capabilities where insurers already run their core operations, Guidewire gives agents the context and controls they need to work with speed and precision on a proven durable foundation.”
“We took a platform approach to AI, so agents run on the same platform as the core, with access to an insurer’s context, data, and workflows,” said Diego Devalle, chief product development officer at Guidewire. “The scalability and observability that already support the core now support agents too. That is why insurers can put them to work on live claims and policies, and why developers can build on the same open foundation.”
Agentic Underwriting
On the underwriting side, Guidewire is pitching speed and precision across the insurance lifecycle. Product managers can move products from intent to market-ready in days with its Advanced Product Designer, using impact analysis, gap detection and explainable recommendations, while actuaries can move sophisticated pricing models to live rates in days through PricingCenter, validating rules, factors and adjustments while keeping portfolio impact visible. Native integration with PolicyCenter connects pricing decisions directly to policy workflows.
Other tools include a Risk Accumulation App, which lets underwriters see concentration and exposure by peril on a map, from a single submission to the whole book. UnderwritingCenter turns raw submissions into clear risk profiles and matches risks to portfolio strategy, with direct integration moving the quote into PolicyCenter. ProNavigator gives underwriters embedded AI for submission summaries, risk insights and next-best actions.
Agentic Claims
Claims teams face similar pressure to act quickly while increasing the precision of each decision, and Guidewire puts agentic capabilities in the flow of work from first notice of loss through adjudication and the broader indemnity supply chain. Concierge lets claimants connect with autonomous agents by voice or chat; the agents capture full loss details and deliver decision-ready claims to ClaimCenter. ProNavigator helps adjusters move from context to decision faster, with answers, suggestions and next-best actions from an embedded AI assistant and Guidewire-developed agents. A Workers’ Compensation Solution links claims work with digital tools and AI to help injured workers recover sooner.
The Platform Underneath
Underneath sits the agentic platform and data layer. An Agentic Framework lets insurers use Guidewire-developed agents or build their own, select the model for each task, enforce guardrails, evaluate behaviour and monitor costs. Developer Assistants help technical teams update code, workflows and rules faster, and DevSpace gives developers preconfigured cloud environments. The Guidewire Data Platform brings policy, claims and billing data together, so agents and predictive models draw on the same data, while Kaleidoscope, a predictive modelling tool tuned to an insurer’s own book, uses federated learning that trains models where each party’s data resides, so records do not move.
One caveat matters for readers: Guidewire notes that certain features may not be available in all regions, and some named features are available for early access customers only or remain on the product roadmap. The announcement is therefore best read as a platform direction with a growing portfolio, rather than a promise that every named capability is generally available today.
Why It Matters
This matters because insurance is a test case for whether agentic AI can survive contact with regulated, high-stakes work. An agent that prices risk or adjudicates a claim cannot afford a mistake, and it must be able to show its work. Guidewire’s bet is that the platform, with its shared data, guardrails and audit trail, is what makes agents deployable in production rather than merely impressive in a pilot, a point echoed by Celent’s Karlyn Carnahan. The same discipline is visible elsewhere, as insurers are given one AI view of a whole portfolio and enterprise platforms are rebuilt around governed agentic action. The risk, as with AI platforms for professional work, is that early access labels blur into availability claims, so insurers should judge this launch on live deployments, not on the conference stage.
Background
Guidewire is the platform property and casualty insurers trust to engage, innovate and grow efficiently, and more than 570 insurers in 44 countries rely on its products, with over 1,700 successful implementation projects supported by its research and development team and consulting partner ecosystem.
Source: Guidewire announcement
0 Comments