OpenAI unveils Dots, a 24-hour agent on GPT-6 Astra

Dots arrives as OpenAI's answer to a field of task-running agents that already includes SpaceXAI's "Grok Bot", Meta's "Meta Muse", and Microsoft's business-focused "Autopilot". What sets the announcement apart is the combination OpenAI describes: a frontier model, GPT-6 Astra; a dedicated cloud computer; and the ability to learn from feedback while working toward a user's goal 24 hours a day. OpenAI says Dots can connect to more than 4,000 apps through its plugin ecosystem and can be reached from ChatGPT, text messages, Slack, or Teams, carrying context across those channels.

Inside OpenAI, the company says Dots is already in use and changing how work gets done. When a bug is reported in Slack, a dot begins investigating; when a new design arrives, a dot turns it into a working app while the team focuses on customer feedback. Outside testers report cases such as a dot noticing a tester had forgotten to bill a client, drafting the invoice, and sending it after approval. OpenAI also plans a limited enterprise preview of specialized dots with dedicated IDs and clear roles, and says it is working with Microsoft to integrate those specialized dots with Agent 365 governance and Microsoft's enterprise identity and access management.

For now, Dots starts with a single dot per user, with future plans for dots to work in teams. The early economics are deliberately open: no additional charge at first, with expanded limits during the first month, and future monthly flat fees for adding dots or increasing their speed and capacity. Whether that structure holds will shape how quickly individual subscribers and workspace administrators move from trying Dots to depending on it.