The shift toward autonomous AI agents has increasingly focused on reducing the friction between background task execution and direct user interaction.
While early generative models operated reactively within standalone chat interfaces, recent agentic architectures run independently in cloud environments to complete multi-step workflows.
As these tools take on longer-term responsibilities, the delivery channel for updates and approvals becomes a critical design consideration. Requiring users to open specialized applications to monitor progress can introduce delays, prompting developers to integrate persistent AI assistants directly into standard communication protocols.
OpenAI recently introduced a proactive texting feature for its autonomous agents, known as dots, enabling direct two-way communication via native messaging applications.
Built as persistent background helpers within the ChatGPT ecosystem, dots operate on isolated cloud infrastructure to handle long-term delegation, research tasks, and tool integrations.
The new integration allows users to interact with their assigned agent through standard SMS platforms, specifically utilizing iMessage on iOS and Rich Communication Services on Android.
This expansion brings voice notes, photos, and standard text input into the agent's context window without requiring the user to remain inside the primary web or mobile application.
The texting interface operates as a dedicated one-to-one channel tied directly to the individual user account.
The agent maintains the same persistent memory, background context, and plugin capabilities across both text messaging and the web platform.
Rather than waiting for a user prompt, the system autonomously sends text notifications when a background task finishes, when monitored conditions change, or when an action requires explicit human confirmation.
For complex tasks involving detailed visual assets or interface approvals, the agent delivers summary updates via text alongside direct links to review the full output within the primary ChatGPT dashboard.
Data handling and system boundaries are structured to isolate the agent within its designated channel.
According to its documentations, setting up direct messaging with an agent begins by navigating to the target dot's profile within the ChatGPT mobile app, desktop application, or a standard web browser, where the newly added 'Texting' configuration option is located. Selecting this option opens the configuration workflow required to bind the agent to a designated mobile device.
To complete the link, users must input their mobile phone number and complete a brief identity verification sequence. Once verified, the system dispatches an initial message containing a custom contact card to save on the device, allowing users to initiate a text conversation and immediately start interacting with the agent through standard messaging channels.
Enabling the feature does not grant the agent or OpenAI access to external text threads, group chats, or general device messaging history.
The routing mechanism utilizes a third-party messaging infrastructure provider, meaning messages are handled via standard carrier networks and are not end-to-end encrypted, though delivery metadata is separated from message content upon receipt.
Managing the integration is handled through the agent settings menu within the main application interface, where users can configure notification preferences or disconnect the phone linkage.
The texting functionality is currently available as an opt-in preview for ChatGPT Pro subscribers located in the U.S.
