OpenAI Tests Dots Personal Agents on the $20 ChatGPT Plus Tier

OpenAI is testing its Dots personal agents with a small group of ChatGPT Plus subscribers, the $20-per-month tier. Engadget reports that Dots had already launched for ChatGPT Pro and higher-tier plans, so the Plus test moves the product down the subscription ladder rather than introducing it.
A standard chatbot waits for a prompt and returns text. Dots are built to act on a user's behalf. They can work with more than 4,000 apps through OpenAI's plugin ecosystem, a set of connectors that let the assistant reach outside services. They can also connect to a user's PC and control it. That second capability is the more consequential one. A chat assistant that drafts an email is one class of software. An agent that can operate the machine on which that email would be sent is another.
Dots were originally powered by GPT-6 Astra, the underlying language model that handles reasoning. They can now also run on GPT-6.1 Terra, giving OpenAI a second backend option for the same agent surface.
Usage terms differ by tier. For Pro and Business Premium subscribers, conversations with Dots do not count toward regular ChatGPT usage limits, provided they do not involve complex tasks. Dots are also coming to all ChatGPT Business Standard users, according to the same report.
On iOS, the newest Dots features include widgets, the home-screen tiles that display information at a glance, and Shortcuts integration, which chains actions together in Apple's automation app. The release also adds iMessage and RCS support, the successor to SMS, and the ability to reply to notifications directly from iOS. Together, those additions place the agent inside the messaging and automation tools many users already rely on.
In my view, the most useful way to read this announcement is through the permissions model. An agent that can drive a desktop and send messages on a user's behalf takes over decisions that used to belong to the individual, such as which accounts it can reach and what it may do without confirmation. Those decisions now depend on product defaults, which matter a great deal. A limited cohort is the normal way to find where those defaults fail.
The broader context here is the tier boundary. The line between a conversation and a complex task determines how much of a subscriber's regular allowance Dots consume, which makes it the variable most likely to shape how heavily people use the feature. Whether that line holds under real workloads is something the test cohort, and then the wider rollout, will have to show.
The upside is concrete. An agent that reaches more than 4,000 apps, operates a PC and responds from a phone's notification shade reduces the number of manual handoffs in a working day. For people who already juggle messaging, calendars and documents, that is the practical promise. The open question is how reliably it delivers once users trust it with more than drafting.


