OpenAI has introduced Dots, a new class of AI agents built on the GPT-6 Astra model. These agents operate within dedicated cloud-based computing environments, allowing them to execute multi-step projects, write and test code, and manage applications autonomously around the clock without constant human supervision.
Each Dot functions on a separate virtual computer equipped with its own browser and access credentials. Users can monitor the agent's progress by accessing this virtual environment. While cloud environments are isolated by default, agents can be granted direct access to a user's local machine if necessary.
Dots integrate with an ecosystem of over 4,000 applications and support interaction via ChatGPT, Slack, Microsoft Teams, and voice communication. The system maintains context across these channels, enabling agents to handle tasks like monitoring user feedback, updating project documentation, or performing ongoing research.
Security is managed through a granular permission system where users define access levels for specific applications. Agents are programmed to pause tasks if security threats are detected. Certain critical actions, such as password changes, remain restricted to human control.
OpenAI is currently testing specialized enterprise versions of Dots for roles in procurement, invoice processing, and customer support. The service is available to ChatGPT Pro and Business Premium subscribers, with a beta version offered to Enterprise clients. Future plans include additional fees for scaling agent capacity.
