Use cases › Customer service & contact center

Travelers: how the agent works and how to build one like it

Summary: Travelers deployed a fully autonomous voice agent for auto first notice of loss (FNOL), the first call a policyholder makes after an accident. According to the source, 85–90% of customers who use it complete their claim through the agent.

Source: https://openai.com/index/travelers/ source read via search extract

Why this use case works

FNOL is a strong agent target because the call follows a predictable script: who, what, where, when, injuries, vehicle details. The data collected is structured, the caller is motivated to finish, and the output (an opened claim) is easy to verify. That combination of fixed intake fields and a clear success event is what to look for in your own voice use cases. (Our analysis.)

Agent pattern: Voice agent

A voice agent pipeline: telephony, speech-to-text, an LLM with tool access, text-to-speech, and a structured hand-off to a human when needed. Small teams usually start with a hosted voice-agent platform rather than wiring the parts themselves.

How a smaller team could replicate it

A smaller insurer, agency or any business with phone intake (towing, restoration, property management maintenance lines) can copy the pattern: script the intake fields, let the agent collect and confirm them, write them into the CRM or claims system, and route anything emotional or ambiguous to a person. (Our analysis, not a claim made by the source.)

Want a rough business case first? Try the AI agent ROI calculator, or ask us to scope it.

Risks and guardrails

Risks: callers in distress, accents and noisy roadside audio can break transcription, and a wrong data field creates claim errors later. Guardrails: read back key facts for confirmation, offer a human transfer at any point, and log the full transcript with each claim. (Our analysis.)

What to measure

Completion rate (the headline metric here), average handle time, percentage of calls transferred to a human, and data-quality errors found downstream.

Related Customer service & contact center deployments

• LendingTree: LendingTree: a multi-agent mortgage assistant (supervisor + workers, LangGraph/MCP). Over 97% of conversations are handled without escalation.
• Maccabi Healthcare: Maccabi Healthcare: voice/chat agents handle info requests, emergency guidance and appointment booking for 2M+ users.
• ENGIE: ENGIE: an Agentforce agent answers billing/EV questions and switches on smart charging through the SAP API.
• Commerzbank "Bene": Commerzbank "Bene": a chatbot that handles 2M+ chats and resolves 70% of them.
• Amdocs: Amdocs: a telco CX agent resolves complex issues end-to-end.

Part of the AgentStack Atlas catalog of 52 sourced agentic AI deployments. Spotted an error? Email agentstackatlas@mail.grokbot.com.