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

# Ask Struct

> Query your production systems in natural language — ask about alerts, logs, metrics, or code

Ask Struct is a natural language interface for investigating your production systems. Ask questions about your alerts, logs, metrics, or code — and Struct will gather context across your connected tools to deliver answers.

<Tip>
  Go to [Ask Struct](https://app.struct.ai/ask) in the web app to try it out.
</Tip>

***

## What You Can Ask

Ask Struct works across your entire connected stack. Some examples of what you can explore:

<CardGroup cols={2}>
  <Card title="Alert Analysis" icon="bell">
    Break down common alerts, identify patterns, and surface recurring issues from your alerting sources.
  </Card>

  <Card title="System Health" icon="heart-pulse">
    Summarize current system health using connected monitors, metrics, and recent error trends.
  </Card>

  <Card title="Issue Investigation" icon="magnifying-glass">
    Investigate a specific error, incident, or alert — Struct cross-references logs, traces, and code to find root cause.
  </Card>

  <Card title="Code Context" icon="code">
    Ask about recent commits, code paths, and how changes may relate to production issues.
  </Card>
</CardGroup>

***

## How It Works

1. Type your question in the Ask Struct input box
2. Struct creates an investigation session and begins gathering context from your connected sources
3. You'll see a structured report with findings, root cause analysis, and suggested next steps
4. Continue the conversation — ask follow-up questions to dive deeper

Each question creates an investigation session that you can revisit at any time from the **Issues** page.

***

## Tips

* **Be specific** — "Investigate the 500 errors in the checkout API" works better than "What's wrong?"
* **Reference services** — Mention specific service names, endpoints, or error messages for more targeted results
* **Connect more sources** — The more observability tools you connect, the more comprehensive the analysis. See [Connections](/connections).

<Note>
  Each investigation uses credits from your plan. See [Usage & Credits](/usage-and-credits) for details.
</Note>