Skip to content

Give your iPhone a goal.

Manina AI sees what’s on your real iPhone screen, works out the next step, and uses a small board to tap, swipe, and type. You describe what you want; Manina AI works through the steps. The board firmware is a one-time step: install the AI iPhone agent firmware. How this iPhone automation AI works covers the screen and the taps.

Start with a free AI service available in the app, or connect your own vision model. We plan to offer a subscription AI service in the future to make getting started even easier.

Start with the app

Already use Claude Code, Codex, DeepSeek Harness, or another agent on your computer? Connect it to your iPhone with Manina AI CLI. Watch the demo, then install the CLI.

You saySet an alarm for 7 am every day.
Manina AI does the tapping.See the screen. Act. Look again.
Your iPhone + a small board. You can stop or step in at any time.

Try a simple goal such as “Open Settings and go to Wi-Fi.” Manina AI reads the screen, takes an action, then looks again before deciding what to do next.

It works with the iPhone you already have. No jailbreak. With the app, a task runs with just your phone and the board after the first setup. With the CLI, your computer’s agent decides the steps and controls the same board over Bluetooth.

Save a task you want to use again. Run saved tasks with Siri or Shortcuts, or keep them within reach through widgets and Control Center.

See what happened. Step in when you need to.

Section titled “See what happened. Step in when you need to.”

Follow a run from the Dynamic Island, stop it there, or return to the app. When Manina AI needs help, it can ask you to continue, skip a step, or take over.

Open a run afterward to replay the steps, with taps and swipes shown on the screenshots. You can also export it for a closer look.

Choosing hardware? Start with the board and cable buying guide.

A Seeed Studio XIAO ESP32S3 plugs into your iPhone. Screen Mirroring lets Manina AI see, and AssistiveTouch lets the board move the pointer and tap. Your chosen vision model decides the next action.

The app and Screen Mirroring receiver currently appear as ESP32 Agent. You’ll see that name during setup.

Need help? See the FAQ or contact support. Read the privacy policy to understand how screenshots and tasks are handled.

Using the earlier computer-based setup? See Git branches.