Cloud-native systems of record (SoRs) have proven critical to every facet of modern carrier operations—and in an AI-driven world, they’re more important than ever.

Across P&C, AI agents are reshaping how carriers work. These autonomous tools now routinely generate preliminary loss estimates based on photos and inspection reports, summarize medical records, draft customer communications, and trigger next steps across the insurance value chain. Their power comes from leveraging probabilistic reasoning to execute nondeterministic tasks such as analysis, summarization, and generation. They thrive on ambiguity, adapt on the fly, and often produce different, uniquely useful outputs from the very same prompt.

But insurance runs on determinism. What’s the premium? Has this claim been paid? Who’s covered? The answers to these questions must be exact, reproducible, and explainable—every time. And in my view, this tension will redefine the core as not just an operational necessity, but also as a strategic differentiator in the age of AI.

Why? Because it’s increasingly clear that SoRs don’t disappear in the age of AI agents—they enable it. With cloud architecture, the modern SoR enforces reliability, security, and auditability while also functioning as a system of insight. It provides the trusted context AI agents need to operate safely, intelligently, and at scale.

Probabilistic Agents Require a Deterministic Backbone

To understand what I mean, let’s look at the bigger picture. AI agents shine at nondeterministic tasks like summarization, generation, heuristic triage, and conversational service. But rating, coverage enforcement, and claims financials must always be precise, reproducible, and explainable.

Modern insurance platforms like Guidewire InsuranceSuite, for example, provide the architectural backbone needed to anchor AI actions with carrier-grade determinism. Unified data models, rigorous workflows, and strict governance ensure compliance, preserve actuarial integrity, and maintain ground truth. Without that deterministic foundation, probabilistic reasoning would quickly spiral into inconsistency and (very) quickly destroy customer trust.

3 Strategic Shifts Defining the Modern ‘Smart Core’

To be clear, this isn’t about merely tapping the core as a kind of headless database. It’s about enabling a system of record that collaborates with, and acts as an intelligent control tower for, AI agents in real time—what some call a “smart” or “cognitive” core. This requires three strategic shifts:

Providing not just storage, but orchestration: The modern core doesn’t just hold data and enforce transactions, it provides real-time decision and orchestration logic.

Moving from rigid workflows to dynamic, event-triggered operations: Work moves when signals fire—telematics pings, document arrivals, weather alerts, customer interactions, etc.—many initiated by AI.

Pivoting from structural compliance to intelligent governance:  Policies, claims, and billing remain fully auditable, but intelligent monitoring becomes continuous and explainable as AI agents act.

Guidewire already supports much of this. And we’re continuously weaving innovative new AI capabilities across the platform, from digital underwriting to product design, claim summarization, partner integrations and beyond.

A System of Insight, Not Just Record

In P&C, a deterministically correct answer is a necessity; a contextually informed one is even better. That’s why the modern core must work not only as a system of record, but as a system of insight that drives accurate, real-time decisioning.

Take InsuranceSuite. It doesn’t just house data but enriches it with embedded analytics. Think property peril intelligence from HazardHub, pricing models from Predict, cyber risk signals from Cyence, and insights from an expanding ecosystem of high-quality sources. These are all embedded directly into workflows to give humans and AI agents structured, trusted context so they can move not only faster, but smarter.

A claim-summary AI agent, for instance, can draft an intelligent claims summary that reflects authoritative coverage, predicted severity, and fraud flags. A digital underwriting agent can recommend endorsements, coverages, and pricing based on live portfolio signals and policy rules. It’s this blend of determinism and contextual insight that enables AI to deliver remarkable business value. This isn’t a case of human versus machine, either. It’s human, machine, and system working together:

• AI agents handle nondeterministic tasks like customer engagement, risk summaries, etc.
• The SoR governs core workflows—underwriting, billing, claims—with deterministic precision
• Humans provide oversight with empathy and expert, ethical judgment

The result: scalable AI operations with carrier control, auditability, and trust.

The Criticality of a Cloud-Native, API-First Platform

AI agents don’t move sequentially—they flood the enterprise as they conduct thousands of concurrent tasks. That makes elastic compute, concurrency, and secure integration non-negotiable. To meet this demand, modern systems of record also require a cloud-native, API-first architecture.

The Guidewire Cloud Platform, for instance, is built for high-volume parallel interactions—exactly what AI agents generate. A robust integration gateway allows agents to read and write securely without undermining data integrity. Elastic compute absorbs spikes in volume. And fine-grained access controls keep compliance intact. It’s a model that ensures AI-driven scale doesn’t overwhelm core systems. Instead, it amplifies their value.

Market Leadership, Proven Ecosystems Are Crucial to Success

As someone who works for the market leader with almost 600 customers worldwide, the industry’s largest partner ecosystem, and a sector-defining insurtech incubator program, I can say with certainty that AI agents don’t kill the SoR—they elevate its importance. It’s a fact that’s already validated in market outcomes.

Working with McKinsey, Guidewire customer Aviva recently harnessed 80 different AI models to shorten liability assessment by 23 days, improve claims routing accuracy by 30%, and reduce customer complaints by 65%—while saving as much as $82 million. This summer, Aviva also began piloting an actuarial AI agent in specialty lines underwriting, tapping live portfolios, third-party risk scores, climate data, and more. And Aviva isn’t alone: A 2025 survey shows AI is already delivering enough savings to help many carriers improve profitability despite rising costs.

As demonstrated with InsuranceSuite and its ecosystem, the modern core isn’t just a system of record. It’s a trusted system of insight that makes AI agents useful, safe, and scalable. The core isn’t fading into irrelevancy in the age of AI agents—it’s stepping into its most important role yet.