> ## Documentation Index
> Fetch the complete documentation index at: https://jam.dev/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# AI apps

> Capture bugs in your AI app with all the context engineers need to debug prompts, model settings, and unpredictable behavior.

Jam captures AI app logs automatically so engineers pinpoint issues fast.

### Custom logs

Use [custom logs](/custom-logs) to log parameters such as temperature, max token length, context window size, or any custom settings you need.

**How it helps.** Spot configuration changes like temperature shifts or new system prompts that cause unexpected behavior. Every bug report includes version numbers, system prompts, and configuration settings.

### Instant Replay

Use [Instant Replay](/instant-replay) to catch unpredictable bugs when a model returns incorrect responses or hallucinations.

**How it helps:**

* **Context drift.** AI apps lose track of earlier context, producing inconsistent or irrelevant outputs. Use Instant Replay to show engineers where the conversation drifted.
* **Error and exception details.** AI bots sometimes fail silently or return a generic error. With Instant Replay you capture the bug the moment it happens. No reproduction required.

*Bonus: with [Jam AI](/jam-ai) you get automatic repro steps, and engineers always get a useful ticket.*

### DevTools

Engineers see full stack traces and console logs in [DevTools](/devtools) to spot issues in your AI app.

**How it helps. Output variability and latency.** When an AI app returns inconsistent responses, console and network logs reveal unexpected API responses or token limits.

### Backend logs

Capture server-side errors with the [Sentry](/sentry) integration so engineers can diagnose rate limits and misconfigured API endpoints.

**How it helps. Backend diagnostics.** When your AI backend fails (HTTP 503 errors, LLM API timeouts), errors log automatically and attach to the bug report.

### Screen recording

Use [video screen recording](/video) to capture the intermediate steps the AI takes while reasoning.

**How it helps. Chain-of-thought debugging.** When answers are incorrect or incomplete, capture the reasoning state so engineers can review the steps the model took.

If we can help you configure Jam for your AI app, [DM us on X](https://x.com/jamdotdev) or [reach out to our team](https://jam.dev/contact-sales).