OpenAI Moves Dot Agents Out of the App and Into Your Texts
You can now text your OpenAI Dot agent directly via iMessage or RCS. The AI assistant now lives exactly where you already message people.
Tools · Source: Engadget
What happened
OpenAI is rolling out a feature that lets you text your Dot agent directly. Instead of opening the ChatGPT app, you can message your agent through iMessage on iOS or RCS on Android. You can send text, photos, and voice memos. The setup requires verifying your phone number inside ChatGPT and saving the contact your Dot sends back.
The agent does not just sit and wait for prompts. It can proactively text you with updates on tasks you assigned. If a task needs your approval, the Dot sends a link so you can review it directly. Memory and context carry over seamlessly between your text thread and the main ChatGPT app. You can even tell the Dot to only text you for urgent updates and keep routine logs in the app.
Right now, this is a limited preview. It is only available to ChatGPT Pro subscribers in the United States. Plus tier users and international users are locked out for now. OpenAI partnered with a company called Linq to handle the messaging infrastructure. They also ran extensive penetration testing to ensure security. The system is strictly one-to-one. Group chats, phone calls, and FaceTime are not supported through the texting interface.
Key facts
- Linq — Partner providing messaging support infrastructure
- US Pro users — Only group with access to the current preview
- GPT-6 Astra — The AI model powering OpenAI Dot agents
- 4,000 — Number of apps Dots can connect to via plugins
- $20 — Monthly cost for Anthropic Pro plan to access Claude Duet
Why it matters
AI is leaving the walled garden of dedicated apps. By moving into native messaging clients, OpenAI reduces the friction of using an AI agent. You do not have to context-switch to give instructions or get updates. The AI lives exactly where you already talk to your team, friends, and family. This normalizes the idea of treating an AI like a human colleague or assistant.
This move sets a new baseline for agent interfaces. Competitors are scrambling. Anthropic just launched Claude Duet, which requires a waitlist and a paid Pro plan. OpenAI is pushing the user experience further by making the agent a literal contact in your phone. Builders need to realize that standalone chat interfaces are commoditizing. Native integration into existing daily workflows is the new competitive moat. If your product requires a separate login and a new tab, you are already losing.
For builders
Build for native channels over web apps
Users want AI where they already spend their time. Forcing users to download a separate app or open a web tab is a losing strategy. Integrate your tools into messaging protocols to win adoption.
Proactive updates are the new standard
Dots do not just answer questions. They message users when a background task finishes or needs approval. If your AI product only responds to prompts, it will feel broken compared to agents that manage their own workflows.
Privacy limits define the feature set
OpenAI explicitly blocked group chats and restricted texting to one-to-one interactions. They also wall off the agent from reading your other texts. Security constraints will dictate how you design collaborative AI features.
My take
Forcing users to open an app is pure friction. OpenAI gets it. By putting the agent in your text messages, they just made every other standalone AI app feel like a chore to use. If you are building an AI product today, you better figure out how to live in the user's existing inbox.
Original reporting: Engadget. This is my rewrite and opinion.