OpenAI has announced ChatGPT Dots, AI agents that can take on tasks and keep working after receiving an initial instruction. Users can give a Dot a task and let it handle the steps involved, rather than providing a separate prompt for every action. Dots can work across connected apps, operate in the background and use their own cloud computers. OpenAI says they can help with software development, research, sales, content creation and everyday tasks, while users can stay in control of what each agent can access.
OpenAI Introduces ChatGPT Dots With Support for 4,000+ Apps
According to the company, each Dot, powered by GPT-6 Astra, gets its own cloud computer and browser and can connect to more than 4,000 apps through OpenAI’s plugin ecosystem. This means a Dot can work across different services while handling a task. Users can also give it another task while it is already working on something else.
For example, OpenAI claims developers can use Dots to track customer feedback, spot common requests, make code changes, test them, and prepare a pull request. Product teams can also use the agents to keep launch material updated when the scope of a project changes. Researchers can use Dots to work with new data, look into unexpected results and update their analysis.
People working in sales can use a Dot to compare what a customer needs with the information available in product documents. It can then help identify what needs testing, prepare a proof of concept and update a proposal when requirements change, according to OpenAI. Content creators can use the agents to go through interview transcripts, find sections that could work as clips, prepare show notes and draft social media posts. For simpler tasks, OpenAI says a Dot can compare food delivery options and work out the final cost before asking the user to approve an order.
Dots can also keep working when a user is no longer in an active ChatGPT conversation. OpenAI says an agent can return with an update, ask a question or seek approval when it needs the user’s input. The same Dot can be accessed through ChatGPT, Slack and Microsoft Teams, so users can continue a task from different services. Voice calls are supported, while text messaging is planned for a future update.
Each Dot works on a separate cloud computer, so setting one up does not automatically give it access to files on the user’s laptop. Users can connect another device if a task requires it and can watch the Dot while it works. On supported websites, Dots can also use saved passwords without exposing those credentials to the model.
Users can choose which apps a Dot can access and set rules for what it can do on their behalf. Actions can be allowed, blocked or made subject to approval. OpenAI says background research is limited to read-only access, so a Dot cannot send messages or make changes to connected apps while carrying out that work. Actions involving accounts or information sharing can also be reviewed through Auto-review, while sensitive tasks such as changing a password remain with the user.
Dots can be stopped if its systems detect malicious instructions or potentially harmful behaviour. The company also warns that the agents can make mistakes, so users should check their work when an error could have serious consequences. Personal ChatGPT users can choose whether their Dot conversations and work are used to improve OpenAI’s models, while Business, Enterprise and Edu workspace content is excluded by default.
OpenAI is also testing specialist Dots for businesses. These agents are being developed for jobs such as procurement, invoice processing, customer support and commercial contracts, and can be given their own identities and access to selected company systems. OpenAI is also working to bring these specialist agents to Microsoft Agent 365.
ChatGPT Dots availability
OpenAI says ChatGPT Dots are rolling out to Pro and Business Premium subscribers in eligible markets, with the first Dot included in both plans at no additional cost. Enterprise users, including those on Edu and Healthcare workspaces, can access the beta when their administrator enables it. The company is also providing extended limits for deeper work during the first month.
According to OpenAI, conversations with Dots will not count towards ChatGPT usage limits. Tasks started or managed through Codex or ChatGPT Work will still count towards those services’ limits. Users currently need to create their first Dot through the ChatGPT desktop app or desktop browser and connect the apps they want it to use. Once set up, users can also access the Dot through the ChatGPT mobile app. OpenAI says it plans to support more Dots and larger workloads in the future.
