OpenAI has introduced Dots, a new set of always-on AI agents within ChatGPT, designed to autonomously carry out user goals across multiple projects without constant input, powered by individual cloud-based GPT-6 Astra instances.
- Persistently running AI agents manage tasks autonomously with dedicated cloud computers.
- Agents communicate across platforms including ChatGPT, Slack, and Teams with shared context.
- Enterprise pilots and proactive AI research features reinforce operational safety and oversight.
What happened
At its DevDay developer conference, OpenAI announced Dots, a new set of always-on AI agents integrated within ChatGPT. These agents can independently handle assigned goals and simultaneously manage multiple projects without needing step-by-step user direction. Each agent is powered by the GPT-6 Astra model and runs on its own cloud computer, which users can monitor at any time to observe ongoing activity.
Beyond passive task execution, agents learn user preferences from feedback and can proactively identify tasks to assist with when idle, operating under strict read-only access until authorized. OpenAI highlighted that these agents can interface with over 4,000 third-party applications through plugins and support seamless communication across platforms such as ChatGPT desktop, web, mobile, Slack, and Teams.
Why it matters
Dots represents a major evolution in AI assistance by enabling continuous, autonomous AI support rather than limiting users to discrete prompt-based interactions. This persistent model allows users to delegate complex workflows to their agents, freeing time and improving productivity. The ability to transfer projects across collaboration tools without re-briefing highlights a focus on improving workflow continuity.
The launch also pits OpenAI directly against competitors like Meta, which recently introduced its Muse personal agent operating from dedicated cloud infrastructure. OpenAI’s emphasis on safety through auto-review checks, monitoring, and controlled permissions addresses growing concerns about AI risk and data integrity in persistent agent deployments.
What to watch next
Dots is rolling out to ChatGPT Pro and Business Premium subscribers today, with Enterprise beta availability contingent on workspace administrator activation. Expansion plans include additional agents, enhanced speed, and increased monthly capacity, though pricing details remain undisclosed. Regional exclusions, such as in the European Economic Area, Switzerland, and the U.K. for Pro users, suggest ongoing regulatory compliance considerations.
On the enterprise front, OpenAI is collaborating with Microsoft to integrate specialist agents within Microsoft’s governance and security frameworks via Agent 365. These specialized agents, provisioned with distinct identities and credentials for specific roles like procurement or invoice processing, will be piloted in controlled customer environments, signaling a move towards tailored AI workforce integration.