OpenAI Introduces 'dots': Always-On Proactive Agents Built on GPT-6 Astra

OpenAI introduced dots, always-on AI agents powered by GPT-6 Astra that run persistently on their own cloud computers and act as proactive digital assistants rather than conversational tools a user has to actively direct.

Details

• Proactive by design: dots use read-only access to connected apps to “look for ways to help in the background,” taking initiative without constant prompting, learning user preferences over time, and sometimes completing work before it’s requested
• How it works: users create a named dot and connect up to 4,000 apps via plugins; a dot can be reached through conversation — text or voice — and maintains context across every channel it’s connected to, working on multiple projects at once using its own browser and cloud computer, with optional access to the user’s own devices if granted
• Example tasks: investigating bugs, converting designs into working apps, coordinating projects, preparing invoices, and handling email marketing or customer support; sensitive actions such as password changes still require human approval
• Availability and pricing: rolling out now to Pro and Business Premium subscribers in eligible markets, with Enterprise users getting beta access; the first dot is included at no extra cost, with additional dots and scaling options to follow
• Vs. standard ChatGPT: dots keep a persistent identity and memory, run on dedicated cloud resources instead of a single chat session, operate continuously across integrated apps, and include built-in approval workflows for account-affecting actions

What happened next

dots was one of the two flagship product announcements at DevDay 2026 alongside GPT-6.1 Sol, and represents OpenAI’s clearest move yet from a chat interface toward an always-on agent model — directly extending the autonomous-computer-use direction previewed in GPT-6 Astra’s release and positioning OpenAI against always-on assistant efforts from rivals.