OpenAI has introduced Dots, a new type of personal AI agent designed to work continuously on tasks with minimal supervision. Announced at DevDay in San Francisco, Dots can interact with more than 4,000 apps through OpenAI's plugin ecosystem, and new integrations with Slack and Microsoft Teams let users message or call their Dot outside ChatGPT. The company says Dots operate on their own cloud computer and browser, and can even be connected to a user's PC to work alongside them.
For businesses, the appeal is productivity: Dots can prepare invoices, integrate lab data into papers, or collect user feedback and draft bug fixes. OpenAI says working with a Dot is as simple as having a conversation, and that Dots carry context across surfaces. However, the company is also positioning safety as a key feature, noting that proactive research tools are restricted to read-only actions by default and that users can set custom rules for when approval is required.
The launch is limited: Pro, Business, Premium, and Enterprise users in select markets get one primary Dot at first, with usage limits in Codex and Work applying as normal. OpenAI says it eventually plans to let users control teams of Dots. The timing is notable, as the company has faced recent criticism over its AI agents, and it is emphasizing safeguards such as using saved passwords without exposing them to the underlying model.