An assistant that keeps working after you close the tab

Most AI tools wait for you. You open a chat, ask for something, get it, and the moment you close the tab the work stops. Anything that happens between conversations is still your job.

Dots are OpenAI's attempt to change that. Launched on September 29, they're what OpenAI calls "always-on agents": a personal assistant inside ChatGPT that keeps working toward your goals when you're not there, notices things that need doing, and brings you finished work to approve. The launch film casts them as friendly little characters, a blue blob in a beret among them, and you give yours a name when you set it up.

How it works

Each dot runs on GPT-6 Astra and has its own cloud computer and browser, so it can browse, build and work through tasks itself. You can open that computer at any time and watch what it's doing. It starts with context from your ChatGPT memory, connects to the apps you choose (OpenAI says over 4,000 through its plugins), and uses Codex for building and coding tasks.

You talk to it in ChatGPT on desktop, web or mobile, by message or by voice call. It also works in Slack and Microsoft Teams, with texting coming next, and it carries the same context across all of them. Start something in ChatGPT, mention it in a Slack channel, and your dot picks up where you left off.

The part that makes it different is what OpenAI calls proactive research. When you're not working with it, your dot looks through your connected apps for ways to help, using read-only access, so it can't send anything or change anything on its own while it's looking.

What it's good for

OpenAI's examples lean toward work that keeps changing under you. A few that will sound familiar to anyone in a creative team:

• Launches: when a product's scope changes, it works out what that means for the story and prepares revised launch materials and docs for your review.
• Decks: it keeps a leadership deck current as new numbers arrive, and flags what still needs figuring out.
• Content: it learns your voice and audience, finds moments to clip from an interview transcript, and drafts show notes and social posts for your approval.

Outside work, one of the launch page's examples is dinner: your dot brings you delivery options with totals, and places the order once you approve.

You stay in charge

Dots come with built-in rules for when to act and when to ask. You can add Custom Rules to allow, require approval for, or block specific actions, and follow everything, including background work, in an Activity View. Anything that could affect your accounts or share information goes through a check called auto-review, and some things, like changing a password, always stay with you. Your own computer stays separate unless you choose to connect it.

Heads up

Access is limited for now. Dots are rolling out to ChatGPT Pro and Business Premium users in eligible markets, and Enterprise, Edu and Healthcare workspaces can try the beta if their admin turns it on. Your first dot is included in those plans, and chatting with it doesn't count toward your ChatGPT limits, though tasks it runs in Codex or ChatGPT Work do. You set it up in the desktop app or a desktop browser before it works on mobile.

It's also brand new, and OpenAI says so plainly: dots can still make mistakes, so review anything that matters. Giving an assistant access to your email, calendar and files is a real decision, and it's worth starting with a small, well-defined job before handing over anything important.

Who it's for

People whose work never quite sits still: creative leads juggling launches, founders, producers and anyone who spends too much of the day keeping things up to date. If you're already on Pro or Business Premium, it costs nothing extra to try. Give it one responsibility you'd happily hand off, and see what it brings back.

Written by Steve Salgado