Skip to main content
Mobilerun Agents are hosted agents you chat with in the dashboard at cloud.mobilerun.ai. Your default agent is Matty. You tell it what you want done, and it works on a real phone on your behalf (Cloud Android, Cloud iOS, or a Personal Phone, Android or iPhone), then reports back with the result. Think of it as an operator for your Mobilerun account. It can set up and manage devices, run apps, handle accounts and logins, scrape data, and build recurring automations, all from a chat window. You don’t write scripts or call an API. You describe a goal, and it figures out the steps.

What it can do

Devices

Create a Cloud Android phone (subscription or Pay as you go), work on your Cloud iOS and Personal Phones, or point you to connecting your own.

Apps and accounts

Install apps, log in, and manage credentials on a device.

Scraping and results

Collect data from an app or page and hand it back as a file or a table.

Routines

Turn a task into a recurring or triggered automation.

Notifications

Tell you in Slack, Teams, your browser, or by webhook when it finishes, fails, or has a question.

Files

Produce downloads, screenshots, and other artifacts from a run.

Integrations

Read Gmail, Calendar, GitHub, Slack, and Teams, and work in Google Drive, Sheets, Docs, Slides, Slack, Teams, and OneDrive.

Recordings

Record the phone screen while it works and play it back in the chat.
It also answers questions about Mobilerun itself: plans and credits, what a job might cost, and where to start if you want to build with the SDK or API instead. For developer questions it points you to the right docs page and example rather than teaching code in the chat.

What it is not

Mobilerun Agents is a chat interface, not the raw automation layer underneath it. It is not the task agent that developers call directly to drive a device step by step. For that, see Agent configuration. It is also not a developer SDK. If you want to build your own integration that talks to the agent over an API, or teach another AI agent to hold this conversation for you, see Mobilerun Agents · chat and the MCP Server page.

How a conversation works

1

You ask

Describe what you want in plain language. No commands, no tool names.
2

It plans

Your agent figures out which device, app, or account the task needs.
3

It may ask back

If something is ambiguous or sensitive, it pauses and asks you a question or asks for approval before continuing.
4

It works on a phone

It runs the task on a real phone, taking as many steps as the goal needs.
5

It presents results

It leads with the outcome: what succeeded, what failed, and any file or link you need.
Your agent never completes a purchase for you. When money is involved, it shows you a card in the dashboard and you click it. See Limits and costs.

Where to go next

Getting started

Your first conversation, first device, and first result.

Devices

Phones, pay as you go, apps, accounts, proxies, numbers, and mailboxes.

Results

How scraped data, diffs, and files are collected and presented.

Routines

Recurring and triggered automation built entirely in chat.

Notifications

Get notified when a run finishes, fails, or has a question for you.

Integrations

Connected services: what your agent can read and where it can write.

Tasks and recordings

Where device work shows up, and how to record the screen.

Human in the loop

How questions and approvals work, and how to walk away and come back.

Memory and skills

What your agent remembers and how to teach it a new app.

Limits and costs

How credits work and what stops a run from getting out of hand.