AI
OpenAI Launches Dots, an AI Assistant to Rival Meta's Muse
OpenAI has unveiled Dots, always-on AI agents that automate tasks across apps and learn user preferences in real time.
The Verge
OpenAI unveiled Dots at its DevDay event this week, introducing a new line of agentic AI assistants designed to compete directly with Meta's recently launched Muse. These always-on helpers operate in the background across integrated apps and services, progressively learning what users value over time.
How Dots Work#
Powered by OpenAI's GPT-6 Astra model, Dots run on their own cloud infrastructure and can interact with web browsers and over 4,000 compatible applications. Users communicate through a messaging-style interface available across ChatGPT on web, desktop, and mobile platforms, or initiate voice conversations directly within the app. Integration with Microsoft Teams and Slack means Dots preserve context from prior interactions when handling requests in those environments; text message support is coming soon.
The interface allows users to check progress on tasks in real time by accessing the Dot's cloud environment. Over time, Dots identify and tackle additional work autonomously—handling chores users haven't explicitly requested yet. Team members can collaborate around a shared Dot within ChatGPT Space, a new shared workspace feature.
Safety and Control#
Unlike Meta's Muse, which has faced concerns about inadvertently sharing user data, Dots come equipped with guardrails. The system ships with "built-in rules for when to act independently," and users can customize settings to prevent specific actions or mandate approval before certain operations run. An "auto-review" mechanism validates that each action aligns with both custom rules and safety standards, determining which sensitive tasks—such as password changes—should be escalated to the user rather than handled by the agent.
Availability and Future Plans#
Starting immediately, Dots are rolling out to ChatGPT Pro and Business Premium subscribers, with one Dot per user currently allowed. OpenAI indicates that multiple Dots per user and scaling options—including speed adjustments and monthly work capacity increases—will arrive later. The company is also piloting "specialist" Dots designed for particular roles within enterprises.

