OpenAI launched Dots, a set of always-on AI agents built into ChatGPT, at its DevDay keynote on Sept. 29, according to The Decoder and BGR.
Each dot runs on OpenAI's GPT-6 Astra model, gets its own cloud computer and browser, and works toward goals a user sets around the clock, the outlets reported. Users can reach a dot through ChatGPT's desktop, mobile or web apps, or through Slack or Microsoft Teams, with voice calls already working and text support planned, according to BGR.
When not actively assigned a task, a dot runs what OpenAI calls proactive research, scanning a user's connected apps with read-only access to figure out what to do next, The Decoder reported. It can act on tasks like following up a bug report or sending a forgotten invoice, but needs explicit approval before sending messages, editing content or taking other sensitive actions, both outlets said.
Dots launched on ChatGPT's Pro 200 and Business Premium plans, with one dot included in the subscription and additional dots costing extra, BGR reported. The launch came three weeks after Meta's Muse, a similar always-on agent that offers free and paid tiers starting at $20 a month, according to The Decoder.
OpenAI is pricing a product only its most expensive subscribers can use to compete with a free rival. That bet says more about what OpenAI expects compute costs to do to its margins than about what users will pay for a digital coworker.