OpenAI unveils Dots, its answer to Meta's Muse assistant

What is OpenAI Dots and how it responds to Meta Muse

On September 29, 2026, OpenAI introduced Dots during its DevDay keynote, positioning the new tool as a direct response to Meta’s recently launched Muse AI. The company described Dots as always-on artificial intelligence helpers that operate in the background across linked applications while continuously adapting to a user’s habits. According to the announcement, each Dot runs on the GPT-6 Astra model and relies on a dedicated cloud computer that can launch a web browser and interact with more than four thousand supported programs.

How to chat or talk with OpenAI Dots using ChatGPT

Users can communicate with a Dot through a chat-style interface that resembles the one offered by Muse, or they can initiate a voice conversation via ChatGPT on the web, desktop, or mobile clients. OpenAI also noted that Dots will be able to connect to Microsoft Teams and Slack, preserving the context of ongoing sessions across devices and apps, and that a text-message capability is slated for release in the near future.

At launch, each account is limited to a single Dot, although OpenAI said it intends to allow multiple agents to run simultaneously later on. When a task is assigned, the Dot sends periodic updates and asks clarifying questions as it works. The company illustrated the concept with a developer who asks a Dot to analyze customer feedback, build and test app updates, and then share a video walkthrough of the changes. A content creator, meanwhile, could task a Dot with scanning an interview transcript, pulling out highlight reels, drafting show notes, and preparing social-media copy.

Monitoring Dot progress and learning from interactions

Progress can be monitored by opening the Dot’s cloud computer, and the system is designed to learn from each interaction. Over time, the assistant tailors its output and may propose additional chores before the user even thinks to request them. OpenAI also highlighted a new collaborative environment called ChatGPT Space, where teammates can jointly interact with a Dot inside the app.

The rollout arrives just weeks after Meta unveiled Muse, a customizable assistant that can browse the web, make purchases, generate documents, and track personal goals. Early adopters of Muse have raised safety concerns, citing an incident where the tool allegedly disclosed a user’s address to a buyer on Facebook Marketplace without consent. In response, OpenAI emphasized that Dots include built-in guidelines governing autonomous behavior, and users can add custom rules to restrict certain actions or require approval before proceeding. An auto-review mechanism checks each proposed action against those rules and safety standards, and it can decide to hand off sensitive tasks-such as password changes-to the human user.

Who can access OpenAI Dots subscription tiers

Starting immediately, Dots are being made available to subscribers of ChatGPT Pro, Business Premium, and Enterprise tiers. OpenAI is also experimenting with a feature that lets organizations create specialist Dots tuned to particular roles within a company. Importantly, conversations with a Dot do not count toward the standard ChatGPT usage limits. Looking ahead, the company said users will eventually be able to add more Dots and scale each one’s capacity by increasing its speed or the amount of work it can handle per month. OpenAI further indicated that a consumer-focused release is planned, though no exact timeline was provided.

For developers, creators, and business teams, the arrival of Dots signals a shift toward AI agents that persistently operate behind the scenes, learn from ongoing activity, and integrate deeply with everyday software tools. The approach aims to reduce manual effort while giving users granular control over what the assistant can do independently. As the technology moves from limited access to broader availability, builders will need to consider how to design workflows that leverage these background agents while respecting privacy and safety safeguards. The coming months will likely reveal how quickly organizations adopt multiple Dots, how they customize the auto-review features, and whether the anticipated consumer rollout reshapes personal productivity in the same way Muse aims to do.

Frequently asked questions

When was OpenAI's Dots announced and what event was it unveiled at?

OpenAI announced Dots on September 29, 2026, during its DevDay keynote. The company positioned Dots as a direct response to Meta’s recently launched Muse AI assistant.

Which AI model powers each Dot and what capabilities does its dedicated cloud computer provide?

Each Dot runs on the GPT-6 Astra model and relies on a dedicated cloud computer that can launch a web browser and interact with more than four thousand supported programs.

How can users interact with a Dot and what integrations are available at launch?

Users can communicate with a Dot via a chat-style interface similar to Muse or through voice conversation using ChatGPT on web, desktop, or mobile clients. At launch, Dots connect to Microsoft Teams and Slack, preserving session context across devices and apps.

What limits exist on the number of Dots per account and how are usage limits handled?

At launch each account is limited to a single Dot, though OpenAI plans to allow multiple agents later. Conversations with a Dot do not count toward the standard ChatGPT usage limits.

Source: The Verge