Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

Video course · 12 chapters · 24 min · certificate

Build 5 Production AI Agents for Insurance Claims

Build a five-agent insurance claims system with Claude Code and LangGraph: grounded synthetic data, typed state, routing with human review, 72 evals, a Gradio dashboard and a reusable skill.

Become a memberAlready a member? Sign in

What you'll learn

  • Plan an agent system with CLAUDE.md and a build spec
  • Create grounded synthetic data from real images
  • Design a typed shared state with Pydantic
  • Build intake, evidence, policy, risk and payout agents
  • Route claims with LangGraph and human review
  • Test the system with evals and ship a dashboard

Chapters

12 chapters · 23:54
  1. 2:24 01Demo Members The finished claims system A tour of the end product: five agents, a dashboard, an audit trail and 72 evaluations.
  2. 1:40 02Stack Members Architecture and tech stack The stack behind the build: Claude Code to write it, LangGraph to orchestrate agents, Pydantic for typed data.
  3. 1:42 03Planning Members Project rules and planning Setting project rules in CLAUDE.md and generating a brief, PRD and data plan with the BMAD method.
  4. 1:13 04Planning Members The build spec Merging the plans into one implementation-ready BUILD_SPEC and preparing grounded synthetic data.
  5. 2:18 05Data Members Grounded synthetic data Generating claim stories, repair estimates, claim histories and customers from image observations.
  6. 2:07 06Practice Members Data work and context rot Why data is most of the job, and how to handle context rot by restarting sessions with data in files.
  7. 2:16 07State Members Outcomes, audit and typed state Writing expected outcomes, auditing the data package, and building the typed ClaimState.
  8. 3:59 08Agents Members Building the five agents Building each agent layer by layer: intake, damage evidence, policy, risk and payout.
  9. 1:37 09Routing Members Routing and audit logs Connecting the agents with LangGraph routing rules that send risky claims to human review.
  10. 1:26 10Testing Members Evals for the whole system Writing 72 evaluations to test routing, policy, risk, payout and image matching.
  11. 1:42 11UI Members The Gradio dashboard Packaging the system in a Gradio dashboard with claim demo, agent timeline, audit trail and evals.
  12. 1:30 12Reuse Members A reusable Claude skill Turning the whole workflow into a reusable Claude Code skill invoked with one command.

Jobs this course suits

Our AI checked this course against 500 jobs; these get the most out of it. Each job links to its learning path.

UnderwritersInsurance Data AnalystsInsurance Claims ProcessorsInsurance ActuariesInsurance Claims ManagersInsurance Risk Analysts