OpenAI has officially shifted its focus from reactive chatbots to proactive agents with the launch of “Dots,” a new category of autonomous digital assistants introduced on September 29, 2026, at the company’s DevDay conference in San Francisco. Powered by the GPT-6 Astra model, Dots are designed to function as persistent entities that can execute complex workflows across thousands of applications without constant human supervision.
The development of Dots is led by Alexander Embiricos, the OpenAI executive and product lead responsible for Dots. These agents evolved from OpenAI’s coding tools (Codex), a lineage central to how the new agents function; while standard AI models communicate through natural language, Dots utilize their coding heritage to interface with software. By generating and executing code in real-time, these agents can navigate APIs and manage data across more than 4,000 integrated applications, including Slack, Microsoft Teams, and Google Drive.
The Shift to Always-On Autonomy
Unlike traditional AI interfaces that terminate a session once a query is answered, each Dot operates on its own dedicated cloud-based computer environment. This architectural choice allows the agent to remain active even when the user’s physical device is powered down or offline. Because the agent lives in the cloud rather than on a local browser tab, it can monitor project status, follow up on emails, or complete multi-hour data processing tasks independently.
OpenAI also introduced “Space,” a dedicated collaborative workspace designed to host these agents alongside human team members. Within a Space, users can assign specific roles to Dots, allowing them to participate in long-term projects rather than just handling one-off requests. This environment serves as a central hub where the agent’s outputs—such as drafted documents or analyzed spreadsheets—are stored and iterated upon by both the AI and the user.
Rollout and Regional Restrictions
At launch, OpenAI is positioning Dots as a premium feature for its paid tiers. The agents are currently available to ChatGPT Pro, Business Premium, and Enterprise subscribers. Under the current structure, Business Premium is priced at $100 per month, while Pro remains at $20 per month. At the initial release stage, Pro subscribers are limited to creating one Dot per user.
The rollout is notably constrained by geography. While the feature is appearing for global users, Pro subscribers located in the European Economic Area (EEA), the United Kingdom, and Switzerland are currently excluded from the Dots rollout. OpenAI has not provided a specific timeline for when these regions might gain access, though such delays are frequently associated with aligning new agentic technologies with regional data privacy and AI governance frameworks.
By evolving the Codex foundation into a consumer-facing agent, OpenAI is betting that the future of productivity lies in “agentic” partners—tools that do not just suggest code or text, but use code to inhabit a functional computer environment and perform work on the user’s behalf.
