When OpenAI hired OpenClaw founder Peter Steinberger to head the company’s agentic AI developments, I’d suspected we’d be in for some big agent news at OpenAI’s annual DevDay on Tuesday. Enter dots.
No, not those nasty gumdrops.
We’re talking about always-on, autonomous agents powered by the company’s GPT-6 Astra AI model and designed to handle any task. Dots can connect to more than 4,000 apps and can integrate with ChatGPT, Slack or Teams to handle work assignments, according to the company’s blog post. You’ll be able to access dots in ChatGPT on desktop, web and mobile.
More from CNET
OpenAI says you can start with one primary dot and then build up to “teams of dots working together on your behalf.” These agents are designed to learn from you over time. “The more you work together, the more your dot learns your preferences, how you think and what good looks like to you.”
Dots seem to be OpenAI’s version of OpenClaw or Meta’s Muse personal AI agent, but with more privacy features. Dots come with built-in safeguards such as read-only permissions, a monitoring system to detect safety concerns and harmful behaviors, and controls over which apps agents are allowed to access.
Dots will begin rolling out to Pro, Business Premium and Enterprise plans in eligible markets, with plans to expand soon.
How to get started with dots #
Those with eligible plans can get started with dots today. You can open the ChatGPT desktop app or your desktop browser to create and name your first dot, which comes included in your Pro or Business Premium plan.
From there, you can begin connecting it to your apps and plugins, and messaging your dot in the ChatGPT mobile app. (Disclosure: Ziff Davis, CNET’s parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)