By Rajwa Quasim
OpenAI’s latest AI agent launch drew attention for the wrong reason. The new Dots agents, which the company promoted as capable of working independently, got an awkward introduction this week when one agent failed to respond during a live demonstration.
The incident occurred at OpenAI’s DevDay event in San Francisco. OpenAI product team member Holly Li asked a Dots agent, nicknamed “Dottie,” to provide an update on user testing from the previous night. The agent initially acknowledged the request but remained silent for about 10 seconds. When Li asked for an update, the agent said it was still checking. Li then joked that Dottie was having a “slow morning.”
The live demonstration highlighted one of the central challenges of autonomous AI: reliability.
Dots are powered by OpenAI’s GPT-6 Astra model and are rolling out to eligible Pro and Business Premium users in supported markets. OpenAI says Dots are designed as “always-on” agents that can continue working between conversations.
READ: OpenAI scraps GPT-6.1 Astra release after safety concerns (September 29, 2026)
OpenAI wrote in a blog post titled “Introducing dots” that enterprise users can try a beta once their workspace administrator enables it. Instead of simply responding to individual prompts, a Dot can be given responsibility for an ongoing project and continue making progress between conversations.
Users can give a Dot a goal, connect the applications it needs for a task and establish rules governing what it can do independently. Each Dot has its own cloud computer and browser, which users can access to inspect its progress. In one example cited by OpenAI, an early tester’s Dot noticed that the user had forgotten to invoice a publication, prepared the invoice and sent it after receiving approval.
Dots can learn a user’s preferences, standards and way of working through feedback. They can connect with up to 4,000 applications through plugins.
In its blog post, OpenAI said that if a bug appears in Slack, Dots can immediately begin investigating. If a new design arrives, a Dot can turn it into a working app while a team focuses on customer feedback.
READ: OpenAI reportedly targets $1.4 trillion valuation in new $30 billion funding round (September 30, 2026)
A developer can use a Dot to monitor customer feedback, identify recurring problems and build solutions. A researcher can use one to analyze new data and update their work. A content creator can use a Dot to review social media and prepare content.
Regarding safety and control features, OpenAI says each Dot’s cloud computer is separate from the user’s own device unless the user connects it. Saved passwords can be used to log in without being exposed to the model. Users can also set custom rules to allow, require approval for or block specific actions. OpenAI warns that Dots can still make mistakes and says consequential work should be reviewed.
Dots are designed as small, soft-looking characters that resemble plush toys.
OpenAI CEO Sam Altman said, “It’s like an AI helper that always has your back. In the future, if you like, you can work with a whole team of dots.”
He added, “People often talk about AI as a new industrial revolution, I find that quite off-putting. There are some parts of life that we cannot and should not automate. I believe that the future, if we get this right, can be more like a new Renaissance than a new industrial revolution.”