OpenAI introduces Dots, always-on digital assistants powered by cloud computers

OpenAI has introduced Dots, a new lineup of always-on digital assistants designed to operate independently across connected applications and the web.

How Dots Work

Powered by the GPT-6 Astra model, Dots run on their own dedicated cloud computer. This setup allows them to interact with a web browser and more than 4,000 supported apps. Users can communicate with a Dot through a text-based interface or by initiating a voice call from ChatGPT on web, desktop, or mobile devices.

Dots can also integrate with workplace communication tools like Microsoft Teams and Slack, carrying over context between different apps and devices. They are designed to handle complex background operations. For instance, a developer might ask a Dot to gather customer feedback, build updates, and test them, while a content creator could have a Dot review interview transcripts, generate show notes, and draft social media posts.

Customization and Safety

To keep operations secure, Dots include built-in rules alongside custom safety configurations. Users can set boundaries that block specific actions or require the assistant to ask for permission before proceeding. An auto-review feature helps the system verify that ongoing actions comply with both safety requirements and user-defined rules.

Availability

Starting immediately, OpenAI is rolling out Dots to users subscribed to ChatGPT Pro, Business Premium, and Enterprise tiers. Initially, users are limited to creating one Dot, though plans are underway to support multiple concurrent agents and higher monthly processing limits.