Airtap iMessage Feature Lets Users Text AI to Control Phone

Airtap introduced an iMessage feature that lets users send one text to instruct an AI agent on a cloud phone to complete tasks such as scrolling TikTok or placing Starbucks orders.

The system runs without the user installing any apps on their own device. It splits into three parts: a language model for understanding instructions, AutoPilot for visual screen control, and a 24-hour cloud phone that stays connected.

Users send the request over iMessage to a US number and the cloud setup executes the steps through simulated taps. Payment and other secure actions still require direct user input.

This setup responds to feedback that agent tools often demand complex app installs and constant device access.

The cloud phone keeps the agent active without draining the user's battery or needing constant Wi-Fi. Airtap states that sensitive operations like payments remain outside automatic execution to reduce risk.

Questions about trust remain central because the agent must access screens and apps on the user's account. Airtap notes that authorization steps for these actions still fall to the person sending the message.

Competitors in agent tools face similar limits around security checks and device permissions. Airtap's approach uses a separate cloud instance to separate the agent from the personal device while maintaining the same visual control method.

A key uncertainty centers on how well the visual control layer handles varied app interfaces over time. The company has not released independent test numbers on error rates across different apps.

Watch for updates on whether Airtap expands supported actions beyond basic tasks or announces third-party reviews of its security model.