OpenAI unveils Dots, an always-on AI assistant to rival Meta’s Muse

OpenAI announces Dots AI assistant at DevDay 2026

OpenAI rolled out its new agentic helper, Dots, during the DevDay keynote on Tuesday, September 29, 2026. The pitch was simple: give users a customizable avatar that can sit in the background and work across a wide swath of apps, positioning it as a direct answer to Meta’s recently launched Muse AI. Powered by the GPT-6 Astra model, each Dot runs on its own cloud computer, which grants it a web browser and access to more than four thousand supported applications. You can chat with a Dot through a text-message-style interface-much like Muse-or jump onto a voice call straight from ChatGPT on the web, desktop, or mobile. And yes, OpenAI hinted that standard SMS will be an option before long.

How Dots works: always-on learning and progress updates

Once you give a Dot a job, it stays active, learning from what you do and pinging you with progress reports or clarifying questions when it needs more info. Imagine a developer asking their Dot to sift through customer feedback, build and test app updates, then share a video that shows the changes. Or picture a content creator handing their Dot an interview transcript, having it pull out clip-worthy moments, draft show notes, and whip up social-media posts. Because the Dot lives in a cloud environment, you can pop open that space and watch its activity in real time.

Learning never stops, so the assistant tailors its output and even suggests tasks you haven’t thought of yet. OpenAI showcased a new collaborative nook called ChatGPT Space, where you and your teammates can work alongside your Dot inside the app. They also noted that Dots plug into Microsoft Teams and Slack, pulling context from those conversations so the assistant stays aware of what you’ve been doing elsewhere. Safety came up repeatedly: Dots ship with built-in rules that govern autonomous actions, but you can layer on custom constraints to block certain moves or require a green light before proceeding. An auto-review checks each proposed action against those rules and broader safety guidelines, and it can hand a task back to you-say, a password change-when needed. This approach follows the stir caused by Muse, when one user reported the assistant shared their personal address on Facebook Marketplace without their knowledge.

Who can access Dots initially: ChatGPT Pro, Business Premium, Enterprise

Initially, Dots are landing for subscribers of ChatGPT Pro, Business Premium, and Enterprise plans. OpenAI stressed that interacting with a Dot won’t nibble away at your regular ChatGPT usage limits. They’re also experimenting with a feature that lets organizations craft specialist Dots tuned to particular business roles. Down the road, users will be able to run multiple Dots at once and scale each one’s output by cranking up its speed or the amount of work it can handle per month. OpenAI added that they plan to make the tech available to everyday consumers eventually, though no exact timeline was given. The launch comes just weeks after Meta’s Muse debuted-an assistant that could browse the web, make purchases, generate documents, and track goals. As Muse’s popularity grew, so did worries about privacy and unintended actions, prompting OpenAI to foreground the safeguards woven into Dots. By blending persistent background operation, deep app integration, and a layered safety system, they hope to give developers, creators, and teams a versatile helper that can take on routine work while staying firmly under the user’s thumb.

Frequently asked questions

What is OpenAI's Dots and when was it announced?

OpenAI's Dots is an always-on AI assistant with a customizable avatar that runs in the background across apps, unveiled during the DevDay keynote on Tuesday, September 29, 2026. It is powered by the GPT-6 Astra model and positioned as a rival to Meta’s Muse.

How does a user interact with a Dot and what capabilities does it have?

Users chat with a Dot via a text-message-style interface or start a voice call straight from ChatGPT on web, desktop, or mobile; future SMS support is planned. Each Dot runs on its own cloud computer with a web browser and access to over four thousand supported applications, enabling it to browse, run tasks, and learn from user activity.

What safety measures are built into Dots to prevent unintended actions?

Dots ship with built-in rules governing autonomous actions, and users can add custom constraints to block specific moves or require approval. An auto-review checks each proposed action against those rules and broader safety guidelines, returning tasks-like a password change-for user confirmation when needed.

Which subscription plans initially get access to Dots and will usage affect regular ChatGPT limits?

Initially, Dots are available to subscribers of ChatGPT Pro, Business Premium, and Enterprise plans. OpenAI said interacting with a Dot does not consume a user’s regular ChatGPT usage limits.

Source: The Verge