Skip to content

Features ​

Everything here works in the app today. Seeing and controlling phones works on every plan. Features marked Pro need the Pro or Team plan, and those marked Team the Team plan; the pricing page compares the plans. To see the app first, take the product tour, or start it with demo phones when no phone is connected.

See and control ​

Every phone on one screen, and any of them in your hands in one click.

A live fleet

Every connected phone is a tile with its screen, its name, its connection, its mode (Auto, Manual or Paused) and anything that needs you. The phones on screen stream live; the rest refresh as pictures, so hundreds of phones stay fast.

One click to take over

Open a phone and work on it with your mouse and keyboard. The page always says whether you are in control, and automation never touches a phone a person is using.

Everything on the phone

Files (drag and drop), app installs and removal, the list of apps, logcat, the clipboard and a terminal with restricted commands.

Organise

Groups (by hand or by rule), pins, filters, saved views and a wallboard for a second screen.

Health at a glance

Battery, charging, temperature, storage, memory, the app in front and the connection, with problems shown in red.

Demo phones

No phone at hand? The first-run setup offers demo phones: simulated phones to watch, control and try every page on. The app restarts with them in a separate folder, your own data stays untouched, and Leave the demo brings it back.

Phones on several computers Team

Phones plugged into other computers join the same fleet through gateways. Create a one-time code on the Gateways page and run the gateway with it on the other computer; its phones then appear with their names, health and history. The gateway connects out to your Hub over an encrypted link, so nothing is opened on that computer, and workflows run on the phone’s own computer and keep going if the link drops.

A phone taken over, with its tabs for files, apps and logs
A phone taken over, with its tabs for files, apps and logs. Shown with simulated phones.

Automate ​

Do a task once, then run it on one phone or every phone, on a schedule, and replay what happened.

Workflows Pro

Build them in a visual editor or as YAML, or record one by doing the task once on a phone. Steps find screen elements by their text and role first, and fall back to image and text recognition on the screen.

Runs on any number of phones Pro

Start a workflow on one phone, a group or every phone that matches; each run has a timeline with screenshots and can be replayed step by step.

Schedules Pro

Run workflows at set times, at intervals or once, with a choice of what happens when a run is missed.

Group actions and broadcast Pro

Install an app, launch it, reboot or take screenshots on many phones at once, or mirror your input from one phone to others.

Safe roll-outs Pro

Publish a new workflow version to a few phones first and roll back if it fails.

Integrations Pro

Start workflows from signed webhooks, for example from your CI pipeline (GitHub’s signature is checked too), and drive the Hub from your own scripts through its API with an API token. The Python and TypeScript SDKs and a CI command with JUnit reports are not published yet.

Replaying a failed workflow run step by step, shown with simulated phones
Replaying a failed run step by step. Shown with simulated phones.
The visual workflow editor, shown with simulated phones
The visual workflow editor. Shown with simulated phones.

AI, with your own model ​

All of the AI features need Pro or Team, and no plan includes AI usage: you bring a key for any AI service, or a model on your computer.

AI tasks Pro

Give a goal for one phone, a group or every phone; the AI works through the same tools you have and you can read every step afterwards.

Agents per phone Pro

An agent with its own instructions, allowed apps, working hours, memory and spending limit.

AI clients Pro

Claude Code, Codex or any MCP client can use your phones through the Hub. Each client gets one scope (every phone, a group or one phone) and an access level, Look only or Look and operate, and every call it makes is in the audit log.

Your choice of model Pro

Any AI service with an API, with your own key — OpenAI, Anthropic, Google, OpenRouter, Groq, Together AI, DeepSeek, Mistral AI, xAI, Fireworks AI, Cerebras or any other OpenAI-compatible service — or a model on your computer or network through Ollama, LM Studio, vLLM or llama.cpp. Budgets per model cap the spend.

Boundaries that hold Pro

AI stays in the apps you allow, asks a person before payments and account-security changes, never types into password fields, and stops at CAPTCHAs, sign-in checks and other human-only steps.

Settings, AI clients: Claude Code allowed to look and operate on the phones in Lab A, and Codex allowed to look only
AI clients, each with its scope and access level. Shown with simulated phones.

Watch over the fleet ​

Problems, failures and trends, in one place.

Attention and alerts Pro

Problems and failed jobs gather in one place; alert rules notify you on the desktop, by email or by webhook.

Dashboards Pro

Success rates, durations, utilisation, app versions and model costs over time.

Screen intelligence Pro

The Screens page reads what every phone shows and groups the phones on the same screen, so the one stuck on an error stands out. It flags a phone whose screen differs from the others at the same workflow step, groups failed runs by step, app version and the screen they ended on, and compares two screens side by side with the changes marked.

The Screens page: simulated phones grouped by the screen they show
The Screens page groups phones by what they show. Shown with simulated phones.

Security and privacy ​

Built to be trusted with real devices: local by default, every action recorded, people in charge.

Local first

The Hub runs on your computer and starts in Local mode: nothing leaves your computer and your network unless you switch to API mode, and the Privacy page lists every destination and what it would receive.

Audit log

Every action on a phone is recorded with who did it; the log is hash-chained so changes can be detected.

Approvals

Rules hold sensitive steps until a person says yes. Built in: AI agents never confirm payments or change account security on their own, and requests that reach more than ten phones wait.

People, roles and scoped permissions Team

More people with their own sign-in and role, and rights for chosen phones, groups, apps or workflows only. API tokens for scripts and CI come with Pro.

Secrets

Passwords are encrypted with a key in your operating system’s keychain, typed on the phone and never shown.

Remote access, if you want it Pro

Turn it on in Settings → Remote access and use the Hub from a browser on another device, with the same pages as the app, over your own network or a VPN. It uses TLS (a certificate the Hub makes, your own, or a reverse proxy), passwords and two-factor sign-in, and remote sessions cannot run shell commands or change secrets and system settings.

The Approvals page with its rules for AI agents and for requests that reach many phones
Approval rules, on every plan. Shown with simulated phones.
The Privacy settings page: Local or API mode, what would leave the computer, and the log of every connection
The Privacy page. Shown with simulated phones.

Use cases ​

  • Testing an app on many phones. Record the test once, run it on every phone after each build, and open the failed runs with their screenshots. More in Android app testing on real phones.
  • A device lab or support desk. Keep an eye on every phone's health, take over the one a colleague asks about, and install builds on a group. More in controlling many Android phones from one PC.
  • Operations on company phones. Run routine checks on a schedule, get an alert when something breaks, and keep a record of who did what.
  • AI agents on real phones. Let Claude Code, Codex or another MCP client use the phones you allow, with approvals for sensitive steps and a record of every call. More in Android MCP server.

Paying for a cloud device farm today? See how BenchFleet compares as a BrowserStack alternative. Working with adb directly? The ADB guide explains the commands, and the free ADB command builder writes them for one phone or every phone.

At CAPTCHAs, sign-in checks and other steps meant for people, the Hub hands the phone to a person, so people stay in charge of every phone.