Testing LLM Fail States with WireMock and Temporal (Live Stream)
Building an AI agent that works when every tool and LLM call succeeds is one thing. But how do you test how it behaves when they don't?
In this live stream, we’ll take an agentic application with a goal-driven loop and explore how to test it beyond the happy path.
Using WireMock, we’ll deliberately introduce failures into tool and LLM calls, giving us repeatable, deterministic failure scenarios that would be difficult or impossible to reproduce reliably against real services.
We’ll then see how the application behaves with Temporal providing durable execution, handling retries, surviving failures, preserving state, and continuing from where it left off.
The result is a practical approach to testing AI applications against the failure scenarios that matter, without waiting for production to discover them.