Travelers automates millions of transactions and cuts claims call center staff by a third with generative AI and Claude

As one of the largest property-casualty insurers in the United States, Travelers processes tens of millions of claims annually across commercial and personal lines. Each claim involves extracting meaning from dense, unstructured data — adjuster notes, medical records, repair estimates, and customer communications — before any decision can be made. First notice of loss calls alone demanded a sizeable dedicated call center workforce to triage and route incoming claims. The cumulative manual effort across underwriting, claims, and operations created a ceiling on scalability and a floor on expenses. Loss adjustment costs remained stubbornly tied to headcount, and the company recognized that sustaining growth while controlling the operating loss ratio required a fundamentally different approach to workflow automation.

Travelers partnered with Anthropic to deploy Claude-powered AI assistants to nearly 10,000 engineers, data scientists, analysts, and product owners — providing personalized tools that integrate directly into existing workflows rather than operating as standalone applications. The company moved rapidly from early experimentation to production deployment, rolling out dozens of generative AI tools across underwriting, claims, and operations. A generative AI voice agent was built specifically to handle first notice of loss calls, replacing a traditionally high-volume manual touchpoint. For eligible claims, straight-through processing was enabled end-to-end, allowing the system to ingest, evaluate, and resolve claims without human intervention. Claude Code was also deployed to accelerate engineering productivity. The deployment model emphasized agentic AI — systems that act autonomously within defined workflows — rather than assistive tools that still require human execution.

The transformation produced measurable outcomes across staffing, operations, and loss economics:

• 33% reduction in claims call center headcount
• 50%+ of claims now eligible for straight-through processing; customers elect that path roughly two-thirds of the time
• 20,000+ employees use AI tools regularly across the enterprise
• Travelers is consolidating four claim call centers into two

These gains translated directly into lower loss adjustment expenses and an improved operating loss ratio. The scale of employee adoption — more than double the initial 10,000-seat technical deployment — indicates that AI tooling extended well beyond engineering teams into frontline operations, underwriting, and customer service functions.

• Agentic AI outperforms assistive AI: Embedding AI as an autonomous actor within workflows — not just a research aid — is what produced material headcount and cost reductions.
• Proprietary data is the moat: Decades of claims history and domain-specific loss data gave Travelers' models a precision advantage that commodity deployments cannot replicate.
• Voice AI at first notice of loss is a high-leverage entry point: Automating the initial inbound claims call removes a major volume driver from call center staffing models.
• Straight-through processing requires end-to-end commitment: Partial automation still requires human handoffs; full eligibility assessment and resolution pipelines are what unlock the efficiency gains.