Free Resource
AI Debugging Checklist
A production-focused debugging sequence for isolating failures in AI features, APIs, prompts, and workflows.
Separate model failure from application failure first.
Capture exact prompts, inputs, outputs, and timestamps.
Verify the full path: browser, API, data, model, and post-processing.
Reproduce reliably
- Save the input that produced the bad output or crash.
- Note the exact environment, account state, and request path.
- Reduce the problem to the smallest failing interaction.
Inspect the chain
- Compare what the UI sent with what the API received.
- Check model parameters, prompt construction, and tool results.
- Inspect retries, timeouts, and fallback behavior in logs.
Close the loop
- Add a regression test or fixture for the failure.
- Update monitoring around the failure mode you found.
- Write down the root cause and the prevention step.
Need help applying this to your system?
Bring the resource, your current stack, and the blocker you want fixed. I'll help you turn it into an implementation plan.