OpenAI launches Dots, its bubbly agentic avatar

OpenAI used its Dev Day event on Tuesday to introduce Dots, a new personal agentic assistant built on GPT-6 Astra. The pitch is big and oddly casual at the same time: these are “always-on agents built to handle everything,” according to the company. Dots are supposed to work on their own, outside any single device or interface, and keep pushing toward user-defined goals with only light supervision.

That’s a different framing from Codex or ChatGPT, which are more clearly tied to a session or a product surface. OpenAI says users can start with a primary dot, give it a name, and shape it over time. The company also says it sees a future where multiple Dots work together on someone’s behalf, which is the sort of sentence that sounds like a product demo and a management memo had a baby.

OpenAI says Dots will be available on Tuesday in ChatGPT for Pro and Business Premium users in eligible markets. They can be launched from Codex or ChatGPT, and users will be able to talk to them through Slack, Teams and other workplace tools, with text message support coming soon. OpenAI’s examples are pointed: a developer could use one to watch customer feedback and apply bug fixes or feature requests, while a scientist could use one to rerun analysis when new experimental data arrives.

There’s also an enterprise angle here. OpenAI says individual Dots can be given identities, credentials and tools through existing systems, and it is already working with Microsoft to plug into Agent 365 security controls. A lot of the underlying behavior already existed in Codex and other agent frameworks. What’s new is the bundle — and the branding, which swaps the usual stern AI look for a bubbly cartoon dot.