Insurance carriers have spent years investing in tech and modernizing their systems – but underwriting remains stubbornly slow. The problem? Manual coordination. Enter: agentic AI.

If you’ve ever watched a submission move through a typical insurance carrier, a surprising reality becomes clear: the process lags, but the systems themselves aren’t the problem. Rating engines work, data sources connect, and workflow platforms route cases correctly.

So why does underwriting still move so slowly? The reason is simple. A submission is received as a set of emails, PDFs, spreadsheets, loss runs, and other supporting documents, presented in both structured and unstructured formats. Someone checks it for completeness, follows up on missing information, assembles the file, pulls external data, runs pricing scenarios, drafts documentation, and routes approvals. At every step, a person is manually carrying the file forward. The bottleneck here isn’t risk assessment: it’s coordination. That’s why cycle times remain stubbornly long despite years of technology investment.

A case for agentic

For carriers facing rising customer expectations, intensifying competition, and increasingly complex risks, the stark reality is that speed and scalability are no longer advantages: they’re imperatives. The next evolution of underwriting isn’t a faster workflow – it’s removing manual orchestration altogether through a new agentic underwriting operating model.

At its core, agentic underwriting industrializes decisions through autonomous coordination. Rather than automating individual tasks, AI agents coordinate the entire process from intake to execution. Most importantly, they determine what happens next and when human involvement is required. This represents a fundamental shift from previous generations of automation.

With agentic underwriting, workflows become self-coordinating, while underwriters engage at defined confidence thresholds where judgment, negotiation, relationship management, and portfolio strategy add genuine value. The result is a transition from assisted underwriting to autonomous coordination.

The changing face of risk

As wildfire exposure expands, flood zones shift, and secondary perils intensify, risk is becoming ever-more dynamic. But traditional underwriting still evaluates many risks only at bind and renewal, leaving significant blind spots between assessments.

At the same time:

  • Digital broker platforms and embedded insurance marketplaces increasingly expect decisions in near real time, making slow responses a competitive disadvantage.
  • Submission volumes continue to rise as distribution channels automate, overwhelming intake processes that still depend heavily on manual effort.
  • Carriers are facing greater pressure to optimize capital deployment amid catastrophe volatility, social inflation, and rising reinsurance costs.
  • Technology itself has matured, with foundation models, document intelligence, and AI agent frameworks now capable of reasoning across complex workflows and coordinating multi-step decisions in new ways.

So, the question is no longer whether underwriting should evolve – but how quickly carriers  can adapt.

An agentic underwriting operation functions less like a chain of handoffs and more like a coordinated production system. Interestingly, some of the highest-return opportunities exist in the least glamorous areas. Intake automation and exposure extraction consume enormous underwriting capacity and often provide the fastest economic returns. More importantly, they create the foundation required for more advanced decisioning capabilities.

This highlights an important reality: agentic underwriting isn’t about replacing underwriters. It’s about freeing them up to spend more time on the decisions that truly require expertise, and less time moving information from one system to another.

The time to transform is now

Historically, underwriting occurs at bind and renewal, while exposures continue to change between those milestones. A property’s wildfire risk can increase, a fleet’s driving performance can deteriorate, or a business’s financial condition can weaken without triggering reassessment. This episodic model is now antiquated.

Agentic AI introduces continuous monitoring, letting underwriting actions be triggered by events rather than calendar dates. The underwriting file becomes a living risk profile that evolves throughout the policy term. In this new era, industrializing underwriting is ultimately an economic decision – not a technology initiative.

Value emerges across key areas expense reduction through autonomous intake and touchless file assembly, loss ratio improvement through more precise risk segmentation and continuous monitoring, and growth through faster quoting, greater distribution reach, and increased underwriting capacity. Capgemini is helping insurers build the foundation for agentic underwriting. We start by consolidating underwriting processes onto a unified workbench and codifying appetite and referral logic, creating the structure AI agents need to operate effectively.

This enables carriers to progressively automate intake, exposure extraction, decisioning, pricing, continuous monitoring, and embedded distribution. Success comes from a phased approach, not a single leap to autonomy. We recommend starting with a high-volume, well-structured line of business, establishing measurable baselines, and delivering quick wins. Carriers that act now will gain deeper risk insights, more precise pricing, and stronger portfolio performance.

Organizations that move decisively won’t simply process risk more efficiently. They’ll understand it better, price it more accurately, and manage it more proactively than ever before. The future of underwriting won’t be defined by faster workflows: it’ll be delivered by intelligent systems that transform underwriting into a lasting competitive advantage.