Plenty of insurers have piloted AI in claims; far fewer have agents they'd trust in production. In this working session, Anthropic's Tom Deaney starts with the fundamentals of building effective agents, then applies them live to a claims workflow, building a working claims agent on stage with Claude Code and talking through each decision: what context the agent needs, where a human stays in the loop, and how you know it's working. It's interactive throughout, so bring questions. To get the most out of it, bring a laptop with Claude Code installed and an Anthropic Console API key and build alongside him. Designed for claims and technology leaders at carriers and TPAs; no coding background needed to follow.