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

# Quickstart

> Get your first automation running in 5 minutes

<Info>
  You'll need a Granite account to follow along. [Sign up free](https://app.getgranite.ai) if you haven't already.
</Info>

## What You'll Build

By the end of this guide, you'll have:

* Created a process in Granite
* Run an AI agent to execute a task
* Watched the execution in real-time with HITL controls

## Prerequisites

* A Granite account
* Chrome or Firefox browser
* 5 minutes

## Step 1: Log In

<Steps>
  <Step title="Open Granite">
    Navigate to [app.getgranite.ai](https://app.getgranite.ai)
  </Step>

  <Step title="Enter Your Email">
    Type your email address and click **Continue**
  </Step>

  <Step title="Check Your Inbox">
    Click the magic link in your email to sign in

    <Tip>
      You can also use **Sign in with Google** for faster access.
    </Tip>
  </Step>
</Steps>

## Step 2: Create Your First Process

<Steps>
  <Step title="Go to Processes">
    Click **Processes** in the sidebar

    <Frame>
      <img src="https://mintcdn.com/granite/KSlfQAiLmsRX4ksG/images/dashboard/processes.png?fit=max&auto=format&n=KSlfQAiLmsRX4ksG&q=85&s=8a8d9693212b09e4f56b9c7736f65598" alt="Processes page" width="1908" height="951" data-path="images/dashboard/processes.png" />
    </Frame>
  </Step>

  <Step title="Create New Process">
    Click the **+ New Process** button
  </Step>

  <Step title="Name Your Process">
    Enter a descriptive name like "Open Notepad and Type Hello"
  </Step>

  <Step title="Describe the Task">
    In the description field, explain what you want to automate:

    ```
    Open Notepad on Windows, type "Hello from Granite!",
    and save the file as hello.txt on the Desktop
    ```
  </Step>
</Steps>

## Step 3: Run the Agent

<Steps>
  <Step title="Click Run">
    Hit the **Run** button to start the agent execution
  </Step>

  <Step title="Watch Live">
    You'll see a live video stream of the Windows desktop
  </Step>

  <Step title="Review Actions">
    The chat panel shows each action the agent takes:

    * Opening applications
    * Clicking elements
    * Typing text
    * Saving files
  </Step>
</Steps>

## Step 4: Use HITL Controls

During execution, you have full control:

<CardGroup cols={2}>
  <Card title="Approve" icon="check">
    Let the current action proceed
  </Card>

  <Card title="Modify" icon="pencil">
    Change the action before it executes
  </Card>

  <Card title="Skip" icon="forward">
    Skip the current action
  </Card>

  <Card title="Cancel" icon="xmark">
    Stop the entire execution
  </Card>
</CardGroup>

<Warning>
  HITL mode pauses execution for your approval on sensitive actions. This keeps you in control of what happens on the target machine.
</Warning>

## Step 5: View Results

After execution completes:

1. Check **Run Logs** to see the full execution history
2. View **screenshots** captured during the run
3. Download the **recording** for review

<Frame>
  <img src="https://mintcdn.com/granite/KSlfQAiLmsRX4ksG/images/dashboard/run-logs.png?fit=max&auto=format&n=KSlfQAiLmsRX4ksG&q=85&s=e35e647296901b1ee47fd41cd0c3e885" alt="Run logs page" width="1908" height="951" data-path="images/dashboard/run-logs.png" />
</Frame>

## What's Next?

<CardGroup cols={2}>
  <Card title="Create RPA Scripts" icon="code" href="/rpa-automation/overview">
    Build repeatable automations with Python
  </Card>

  <Card title="Set Up API Endpoints" icon="globe" href="/public-api/overview">
    Trigger automations from external systems
  </Card>

  <Card title="Learn HITL" icon="hand" href="/agent-execution/hitl-workflows">
    Master human-in-the-loop workflows
  </Card>

  <Card title="Explore Analytics" icon="chart-line" href="/dashboard/analytics">
    Track your automation performance
  </Card>
</CardGroup>

## Troubleshooting

<AccordionGroup>
  <Accordion title="I didn't receive the magic link email">
    * Check your spam/junk folder
    * Make sure you entered the correct email
    * Try again or use Google OAuth instead
  </Accordion>

  <Accordion title="The agent isn't executing">
    * Ensure you have at least one driver machine connected
    * Check the **Driver Management** page for machine status
    * Verify the driver is online (green status)
  </Accordion>

  <Accordion title="Video stream isn't loading">
    * Check your browser allows WebRTC connections
    * Try refreshing the page
    * Ensure you're not behind a restrictive firewall
  </Accordion>
</AccordionGroup>
