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

# Get Started with BeyondGuard: Deploy Your First Guard

> Connect BeyondGuard to your AI pipeline, deploy your first security guard, and start intercepting threats in four straightforward steps.

This guide walks you through everything you need to go from zero AI-specific security controls to an active guard monitoring your LLM interactions. By the end, you will have BeyondGuard connected to your AI pipeline, Prompt Guard enabled on a project, and your first threat events appearing in the dashboard. The entire process takes under thirty minutes for a standard LLM integration.

<Steps>
  <Step title="Request Access">
    BeyondGuard is deployed for enterprise teams through a guided onboarding process.

    1. Visit [beyondguard.ai](https://beyondguard.ai) and click **Request Access**.
    2. Fill out the access form with your organization name, deployment environment, and the AI use cases you want to protect.
    3. The BeyondGuard team will provision your organization account and send your credentials — including your **API key** and **Control Plane URL** — to the email address you provided.

    Once you receive your credentials, keep your API key secure. Treat it like any other long-lived secret: store it in your secrets manager, not in source code.
  </Step>

  <Step title="Connect Your AI Stack">
    Add BeyondGuard to your AI pipeline by installing the SDK or calling the REST API directly. BeyondGuard wraps your existing LLM calls — no changes to your model configuration, hosting, or prompt templates are required.

    Install the SDK for your environment:

    ```bash theme={null}
    # Python
    pip install beyondguard

    # TypeScript / Node.js
    npm install beyondguard
    ```

    Then wrap your LLM call with BeyondGuard protection:

    ```python theme={null}
    import beyondguard

    client = beyondguard.Client(api_key="YOUR_API_KEY")

    # Wrap your LLM call with BeyondGuard protection
    response = client.protect(
        prompt=user_input,
        guard="prompt",
        project="my-ai-app"
    )
    ```

    ```typescript theme={null}
    import BeyondGuard from "beyondguard";

    const client = new BeyondGuard({ apiKey: "YOUR_API_KEY" });

    // Wrap your LLM call with BeyondGuard protection
    const response = await client.protect({
      prompt: userInput,
      guard: "prompt",
      project: "my-ai-app",
    });
    ```

    The `protect` call intercepts the prompt before it reaches your model, evaluates it against your active guards and policies, and returns either a cleared result or a threat event depending on what it finds. Your downstream model call only executes if BeyondGuard clears the input.
  </Step>

  <Step title="Deploy Your First Guard">
    With your pipeline connected, activate Prompt Guard from the Control Plane dashboard.

    1. Log in to your Control Plane at the URL provided in your onboarding email.
    2. Navigate to **Projects** and select or create the project that corresponds to your AI application.
    3. Open the **Guards** tab and locate **Prompt Guard**.
    4. Toggle Prompt Guard to **Enabled** and choose an initial enforcement mode. If you are new to BeyondGuard, select **Observation** mode to collect threat data without blocking traffic — you can switch to **Enforcement** mode once you have reviewed your first set of threat events.
    5. Save your configuration. Prompt Guard is now active on your project.

    At this point, every prompt that passes through your `client.protect()` call is being evaluated for jailbreaks, prompt injection, scope violations, and system prompt leakage.
  </Step>

  <Step title="Review Threat Events">
    Navigate to the **Threat Events** dashboard in the Control Plane to see what BeyondGuard has detected.

    Each event entry shows:

    * The **threat category** (for example, prompt injection or jailbreak attempt)
    * The **guard** that flagged it (Prompt Guard, in this case)
    * The **project** and **timestamp**
    * The **action taken** — blocked, flagged, or passed in observation mode
    * A **confidence score** and the reasoning behind the decision

    Use the Threat Events view to understand your threat landscape before tightening policies. Filter by guard, by threat type, or by time window to identify patterns. When you are ready to move from observation to active blocking, return to the Guards tab and switch Prompt Guard to **Enforcement** mode.
  </Step>
</Steps>

<Tip>
  Start with **Prompt Guard** before enabling other guards. It covers the widest range of input-layer threats and gives you the fastest signal about your AI application's exposure. Once you have a baseline in the Threat Events dashboard, add **Output Guard** next to close the output-delivery layer — then layer in Agent Guard, MCP Guard, and Data Guard as your deployment grows.
</Tip>

<Note>
  You do not need to reach L5 (Continuous Governance) on day one. BeyondGuard's maturity model is designed to let you start at L1 (Visibility) — simply collecting threat data — and advance to higher levels as your team builds confidence and your policies mature. Sustainable governance is built incrementally, not deployed all at once.
</Note>