OpenAI has introduced Dots, a new generation of personal AI agents designed to pursue user-defined goals continuously with limited supervision. Powered by GPT-6 Astra, the feature is positioned as an always-on assistant that can operate beyond a single device or interface.
Available through ChatGPT and Codex for eligible Pro and Business Premium users, Dots can be personalised with names, roles and dedicated responsibilities. Users can assign a primary Dot to manage recurring work, while future plans include coordinated teams of specialised agents working together.
From prompts to persistent workflows
Unlike conventional chat-based assistants that respond to individual requests, Dots are intended to handle longer-term workflows in the background. A software team, for example, could use a Dot to review customer feedback, identify priorities and support bug-fix planning. In research environments, an agent could revisit analyses as new experimental data becomes available.
Dots can also be configured with specific identities, credentials and tools through existing workplace systems. OpenAI is developing integration with Microsoft Agent 365 security controls, aiming to support enterprise-level management of these autonomous workflows.
Users will be able to communicate with Dots through collaboration platforms including Slack and Teams, with text messaging functionality planned for a later stage. The interface presents agents as simple animated characters, offering a more approachable visual identity for advanced AI tools.
By combining task automation, persistent context and specialist roles, Dots signals a shift toward AI systems that may become active participants in everyday creative, technical and scientific work.
