Thirteen modules from a simple agent loop to a durable production system. State, checkpoints, idempotency, failure taxonomies, human-in-the-loop, context engineering, observability and evals.https://learn.theairuntime.com/en-ushttps://learn.theairuntime.com/courses/long-running-agents/00-mental-model/https://learn.theairuntime.com/courses/long-running-agents/00-mental-model/Understand what makes long-running agent execution fundamentally different from an ordinary LLM request.https://learn.theairuntime.com/courses/long-running-agents/01-naive-agent/https://learn.theairuntime.com/courses/long-running-agents/01-naive-agent/Build something that works, in plain Python, before introducing any framework.Failure Lab 01: Kill the Agenthttps://learn.theairuntime.com/courses/long-running-agents/02-state/https://learn.theairuntime.com/courses/long-running-agents/02-state/Separate conversation, execution state, artifacts, and external reality, and make state represent evidence, not claims.Failure Lab 02: False Completionhttps://learn.theairuntime.com/courses/long-running-agents/03-durable-execution/https://learn.theairuntime.com/courses/long-running-agents/03-durable-execution/Make the agent survive application failure using LangGraph with a PostgreSQL checkpointer.Failure Lab 03: Pull the Plughttps://learn.theairuntime.com/courses/long-running-agents/04-side-effects/https://learn.theairuntime.com/courses/long-running-agents/04-side-effects/Learn the concept most agent tutorials skip: a checkpoint does not guarantee an external operation happened exactly once.Failure Lab 04: The Duplicate Reporthttps://learn.theairuntime.com/courses/long-running-agents/05-work-ownership/https://learn.theairuntime.com/courses/long-running-agents/05-work-ownership/Guarantee that exactly one worker executes a run at a time, and that a run whose worker dies is reclaimed automatically.Failure Lab 05: Two Workers, One Runhttps://learn.theairuntime.com/courses/long-running-agents/06-failure-handling/https://learn.theairuntime.com/courses/long-running-agents/06-failure-handling/Stop treating every exception as retry(3). Build a failure taxonomy and route each class differently.Failure Lab 06: The Failure Matrixhttps://learn.theairuntime.com/courses/long-running-agents/07-human-control/https://learn.theairuntime.com/courses/long-running-agents/07-human-control/Show how a workflow can stop for hours without keeping a worker occupied, and resume with full state.Failure Lab 07: The Three-Hour Approvalhttps://learn.theairuntime.com/courses/long-running-agents/08-context/https://learn.theairuntime.com/courses/long-running-agents/08-context/Treat execution state and model context as separate engineering surfaces; keep the model's working set small, evidenced, and reconstructible.Failure Lab 08: Context Rothttps://learn.theairuntime.com/courses/long-running-agents/09-observability/https://learn.theairuntime.com/courses/long-running-agents/09-observability/Make every run explainable to a human in under a minute, from the dashboard, from the trace, and from the database.Failure Lab 09: The Silent Runhttps://learn.theairuntime.com/courses/long-running-agents/10-security/https://learn.theairuntime.com/courses/long-running-agents/10-security/Bound what untrusted content can make the agent do: tool permissions, SSRF controls, prompt-injection boundaries, and tenant isolation.Failure Lab 10: The Hostile Vendorhttps://learn.theairuntime.com/courses/long-running-agents/11-evaluation/https://learn.theairuntime.com/courses/long-running-agents/11-evaluation/Make evals the development loop: error analysis on real traces, binary criteria, a validated judge, and a regression gate that proves version B beats version A.Failure Lab 11: The Judge That Agreed With Everyonehttps://learn.theairuntime.com/courses/long-running-agents/12-deep-agents/https://learn.theairuntime.com/courses/long-running-agents/12-deep-agents/Rebuild the agent on a higher-level harness, verify every invariant still holds, and learn to tell framework features from guarantees.Failure Lab 12: Replace the Harnesshttps://learn.theairuntime.com/courses/long-running-agents/13-industry-architectures/https://learn.theairuntime.com/courses/long-running-agents/13-industry-architectures/Recognise the same six primitives you built inside every credible production long-running agent, and spot the marketing that hides their absence.https://learn.theairuntime.com/courses/long-running-agents/14-deploy/https://learn.theairuntime.com/courses/long-running-agents/14-deploy/Put the agent on a public URL, for free, and prove that a redeploy mid-run does not lose work.Failure Lab 14: The Redeployhttps://learn.theairuntime.com/courses/long-running-agents/15-production-gauntlet/https://learn.theairuntime.com/courses/long-running-agents/15-production-gauntlet/Run every failure at once against your deployed agent, measure it, and publish a reliability report anyone can reproduce.Failure Lab 15: All of them, at once