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

# Getting started

> Go from your first conversation with your agent to your first result on a device.

This page walks you from signing in to reading your first result. It takes a few minutes.

**1. Sign in**

Create a free account at [cloud.mobilerun.ai](https://cloud.mobilerun.ai). No base plan is
required, and you don't need to add a device before you start chatting.

**2. Say hello**

Open a new chat and say what you want to do, or just say hi. If you don't have a device yet,
your agent asks about that before it asks anything else. It only asks what it can't already
see from your account, so it won't repeat a question you've already answered.

**3. Get a phone**

Every task runs on a device, so your agent helps you get one if you don't have one.

* **Personal Phone** connects a phone you already own. Your agent points you to the connect guide; eligible accounts can claim one Personal Phone slot free.
* **Cloud Android** is a persistent virtual device on a monthly subscription. It requires an
  active dedicated Mobilerun proxy.
* **Pay as you go** is a Cloud Android phone that draws on your plan's device minutes. You can stop it to pause billing and resume it later.

Your agent creates a Pay as you go phone when you ask for one explicitly, and checks with you
first when your request doesn't clearly accept per-minute billing. See [Devices](https://docs.mobilerun.ai/agents/devices.md) and [Device Types](https://docs.mobilerun.ai/device-types.md) for the full
comparison.

**4. Give it a real task**

Describe a concrete goal instead of a vague one. "Automate my phone" is too broad; "open
Settings and tell me the Android version" is something it can just do. See the example prompts
below.

**5. Read the result**

Your agent leads with the outcome, not a play by play. It states what succeeded or failed,
gives you the file, link, or count you asked for, and offers a next step only when a decision
is actually needed.

**6. Answer if it asks**

Mid-task, your agent sometimes pauses to ask a clarifying question or to get your approval
before a sensitive action, like resetting a device. Answer it and the task continues. See
[Human in the loop](https://docs.mobilerun.ai/agents/human-in-the-loop.md) for how these work and what happens if you
walk away.

## Prompts to try

Copy any of these into your first chat, adjusted to your own app or site.

> Open Settings and tell me the Android version.

> Log into this app with my account and tell me my current balance.

> Scrape all attendees of this Luma event and give me a CSV.

> Check my inbox app and summarize anything unread from the last 24 hours.

> Take a screenshot of the home screen so I can see what's installed.

> Check this page every morning and tell me if the price drops.

> [!TIP]
>
> Be specific about the app, the data you want, and the format. A precise goal takes fewer steps
> to complete, and fewer steps means a cheaper run. See [Limits and costs](https://docs.mobilerun.ai/agents/limits-and-costs.md).

## Related

> [card] [Overview](https://docs.mobilerun.ai/agents/overview.md)
>
> What your agent is and what it can do.

> [card] [Devices](https://docs.mobilerun.ai/agents/devices.md)
>
> Cloud Android phones, pay as you go, apps, and accounts in more depth.

> [card] [Human in the loop](https://docs.mobilerun.ai/agents/human-in-the-loop.md)
>
> Questions, approvals, and coming back to a paused run.

> [card] [Cloud Quickstart](https://docs.mobilerun.ai/cloud/quickstart.md)
>
> Run a task from the API instead of chat.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.