Travelers

Travelers, one of the largest commercial insurers in the United States with roughly 33,000 employees, faced a structural data problem common to legacy carriers: decades of underwriting, claims, billing, and distribution records stored across siloed systems, with the majority of that institutional knowledge locked in unstructured formats — adjuster notes, policy documents, call transcripts. This fragmentation slowed claims handling at exactly the moments customers need speed most, such as immediately after an auto accident. Manual intake processes created friction, extended cycle times, and strained contact center capacity, while competitors began experimenting with digital-first claim experiences.

Travelers partnered with OpenAI to develop an AI Claim Assistant — an agentic voice service built on generative AI that handles inbound customer calls for auto damage claims without requiring human agents for routine interactions. The assistant guides policyholders through the entire initial filing workflow: confirming coverage details, answering common policy questions, helping customers decide whether to file, and directing them to complete follow-on steps — submitting photos of vehicle damage, scheduling repairs at approved shops, and arranging rental cars — through Travelers' mobile app. The agentic design means the system orchestrates multiple actions within a single conversation rather than simply answering questions, representing a meaningful step beyond conventional IVR or chatbot deployments. A parallel partnership with Anthropic supports the data science and engineering teams building internal AI tooling.

The AI Claim Assistant is among the earliest customer-facing generative AI deployments in the commercial insurance sector. Internally, Travelers' AI transformation is operating at enterprise scale:

• 20,000+ of 33,000 employees use AI tools on a regular basis
• Dozens of generative AI tools are already in production across the organization
• 10,000 analysts and engineers are supported through the separate Anthropic partnership

These figures indicate that the Claim Assistant is not an isolated pilot but part of a coordinated, company-wide AI strategy. The breadth of internal adoption suggests the infrastructure, governance, and change management needed to sustain customer-facing agentic deployments are already in place.

• Unstructured legacy data is both the biggest barrier and the biggest opportunity for incumbent insurers — investment in data readiness precedes effective AI deployment.
• Agentic AI that completes multi-step workflows (file, photo upload, repair booking) delivers more customer value than conversational AI limited to Q&A.
• Enterprise-wide internal AI adoption (60%+ of workforce) creates the operational confidence needed to extend AI to customer-facing, high-stakes interactions like claims.
• Dual-vendor AI strategy (OpenAI for customer experience, Anthropic for internal tooling) allows firms to match model capabilities to use case requirements.
• Starting with auto damage claims — high volume, relatively standardized — provides a lower-risk entry point before expanding agentic AI to more complex commercial lines.