Dots, an always on AI agent; GPT 6.
OpenAI used its annual Dev Day conference on Tuesday to argue that AI agents have moved past "trivial" handoffs into delegated "ambitious pieces of work," and rolled out the products to back that claim. The centerpiece is Dots, an always-on, cloud-hosted agent reachable through ChatGPT, Slack, and Teams and powered by the GPT‑6 Astra model. Each Dot runs on its own cloud computer with access to more than 4,000 apps, the company said.
Underneath sit three supporting pieces: GPT‑6.1 Sol, a cheaper and faster sibling model; a cloud-hosted Codex; and a new Agents API for organizations that want to build similar agents themselves. Sam Altman framed the pitch as letting users "delegate ambitious pieces of work" to agents reachable by phone call, a boundary the InfoWorld writeup flags as exactly the kind of claim enterprises will pressure-test.
Rollout is narrower than the marketing language suggests. Dots is available to Pro and Business Premium users in eligible markets; Enterprise, Edu, and Healthcare can try the beta only when a workspace administrator enables it. Integration with Microsoft Agent 365 is still in progress. OpenAI's safety writeup says proactive research runs in read-only mode with code-enforced restrictions against messaging, changing connected-app content, or taking over a browser or desktop.
Independent keynote coverage from Simon Willison's live blog corroborates the announcements but describes demonstrations rather than sustained enterprise operating results. No verified enterprise productivity, reliability, or adoption numbers have been published.