ChatGPT dots texting: how to set it up on iMessage or RCS

ChatGPT dots texting lets you message your dot from your phone's regular texting app, over iMessage on an iPhone or RCS on Android. To turn it on, open your dot's profile in ChatGPT, pick Texting, and verify your phone number. It is a beta for ChatGPT Pro users in the US.

The announcement came from @ChatGPT on October 28: "By popular request, you can now text your dot." The post says you can hand the dot a task, or something to keep an eye on, using text, photos or voice notes, and it will text you on its own when there is a relevant update. A follow-up post in the same thread from @ChatGPT covers setup and privacy, and says the feature "doesn't give your dot or OpenAI access to your other text conversations."

If dots are new to you, we covered the launch at DevDay in what OpenAI's always-on dots do. The short version: a dot is an agent that keeps running after you close the app. Until now you had to open ChatGPT to see what it was up to. With texting, it comes to you.

How to set up texting for your dot

OpenAI's release note and its channels help page give the steps:

1. Open your dot's profile in the ChatGPT mobile app, the desktop app, or ChatGPT in a desktop browser.
2. Select Texting, enter your phone number and complete the verification.
3. Save the contact card your dot sends you.
4. Text it. On an iPhone, send from the number you verified. On Android, turn RCS on in your messaging app first.

That's the whole setup. Your dot keeps the same context and plugins whether you reach it in ChatGPT or by text, so anything it saves to memory over text is there when you open the app later. If you already connected plugins like the ones in our ChatGPT iCloud Mail and Calendar setup guide, your dot can use them while you text it.

What you can do over text

You can also tell the dot which channel to use. OpenAI's own example is to text you when something needs your attention and keep routine updates in ChatGPT. When the dot needs your approval for a step, it texts you a link and you review the request inside ChatGPT.

A few prompts to start with:

• "Watch my inbox for a reply from the landlord and text me when it lands."
• A photo of a supplier invoice with "Add this to my bills list and remind me Friday."
• A voice note on the drive home: "Draft a follow-up to today's estimate and hold it for my approval."

What to watch before you rely on it

Engadget reports that OpenAI works with a messaging company called Linq to carry the texts and that it ran penetration tests on the setup. Even so, your texts to the dot pass through a third party on the way to OpenAI, so treat the thread the way you would treat your ChatGPT history. Keep card numbers and passwords out of it.

My read: it's a small feature, and I think it's the one that gets people to actually use dots. An agent that runs all day is only useful if you see what it reports, and I don't open a separate dashboard to check on anything. A text shows up where I already look. I also like that approvals stay in the app. You see the whole request before you say yes, instead of firing off "ok" from a stoplight.

The limits are real, though. It's Pro-only and US-only. There are no group threads, so a team can't share one dot in a chat, and you can't call it the way you'd call a contact. To compare it with Anthropic's proactive agent, see how Claude Duet works. Microsoft's take, from late September, is in our Copilot Autopilot post.

What it means for a small business

Texting an agent works a lot like texting an employee, and the same ground rules apply. Decide what it can do on its own and what needs a yes from you, then tell it which updates are worth a text. Write that down before you switch texting on and you'll get the help without a stream of pings at dinner.

If you want help writing those rules for your shop, our free process audit maps where your hours go and which tasks an agent could safely take over.