OpenAI launches dots AI assistants for ChatGPT users

OpenAI has launched dots, a new AI assistant product for ChatGPT users. The rollout begins with Pro, Business Premium and Enterprise customers in eligible markets.

Dots are persistent software agents that can carry out tasks on a user's behalf across multiple applications and communication channels. Each dot has its own cloud computer, can connect to more than 4,000 apps through plugins, and can be reached through ChatGPT, Slack and Microsoft Teams.

The product marks a broader push beyond chatbot interactions into software that can manage work over time with limited supervision. Dots can learn from user feedback, track goals and preferences, and continue working in the background while users focus on other tasks.

Users start with a primary dot that they can name and customise. Over time, OpenAI expects to support teams of dots working together on behalf of an individual or organisation.

Workflows

OpenAI outlined a range of day-to-day uses for dots. For developers, a dot could monitor customer feedback, identify recurring bug reports or feature requests, build and test smaller fixes, and present pull requests for review. In product marketing, a dot could update launch materials when product scope changes and prepare revised assets and documents for approval.

Other scenarios include scientific research, where a dot could rerun analysis as new data arrives and update figures for a paper, and sales, where it could review customer requirements, help assemble proofs of concept and revise proposals as technical conditions change. OpenAI also pointed to media and creator work, where dots could identify clips from interview transcripts, draft notes and prepare social posts.

Users can inspect a dot's work by opening its cloud computer. Dots can also be given permission to use a person's laptop directly, placing them alongside existing workflows rather than limiting them to a separate environment.

Channels and control

Dots can be contacted through desktop, web and mobile versions of ChatGPT, and can send progress updates, questions and requests for decisions back to users. The product is also available in Slack and Teams, allowing work to move between direct chat and broader team discussions.

OpenAI placed particular emphasis on oversight. Users choose which apps a dot can access, manage permissions through existing ChatGPT controls, and can set custom rules to allow actions, require approval or block them. Sensitive actions, such as changing a password, remain with the user.

Dots can carry out “proactive research” when users are not actively engaging with them, but this background activity is restricted to read-only tools. That means a dot cannot send messages, alter app content or control a browser or computer unless specific permission has been granted.

Each dot runs on a separate cloud computer, while a user's own device remains separate unless explicitly connected. Saved passwords can be used for supported sites without exposing those credentials to the model.

Enterprise focus

Alongside general-purpose dots, OpenAI also previewed specialist dots aimed at businesses. These are intended to handle narrower responsibilities such as access management, IT-provisioned hardware tasks and deeper integration with internal systems of record.

For these uses, each specialist dot would have its own identity, credentials and access rights set by the employer. OpenAI is starting with focused enterprise pilots, with its engineering teams working directly with customers to define each dot's remit, tools and approval processes.

OpenAI said the effort builds on internal testing across procurement, invoice processing, email marketing, customer support and commercial contracting. It is also working with Microsoft to integrate specialist dots with governance and security controls in Microsoft Agent 365.

On data handling, content from ChatGPT Business, Enterprise and Edu workspaces is not used by default to improve OpenAI's models. For personal ChatGPT plans, users can choose whether conversations and work from dots are used for model improvement. OpenAI added that it does not train directly on proactive research activity or a dot's internal notes.

The initial release includes one dot at no extra cost for Pro and Business Premium subscribers. Conversations with a dot do not count toward standard ChatGPT usage limits, although tasks handed off into Codex or ChatGPT Work are counted in the usual way.

“Dots can still make mistakes, so always review consequential work,” OpenAI said.