OpenAI Unveils 'Dots' Autonomous AI Agents Powered by New GPT-6 Astra Model

OpenAI on Tuesday announced the launch of Dots, a new class of autonomous AI agents designed to independently manage and execute complex tasks for users around the clock. The new agents are powered by the company's next-generation model, GPT-6 Astra, and are being rolled out to select subscribers.

Key Features and Capabilities

According to the announcement, each Dot operates on its own dedicated cloud computer, complete with a browser and the ability to connect to more than 4,000 apps through OpenAI's existing plugin ecosystem. The agents are designed to learn a user's preferences and work styles over time through feedback.

Users can interact with their Dot via messaging or voice calls through platforms like ChatGPT, Slack, or Microsoft Teams. OpenAI outlined several potential use cases for the agents, including:

• Investigating software bugs
• Converting design files into functional applications
• Preparing and managing invoices
• Overseeing project management workflows

Phased Rollout and Enterprise Focus

Ad

OpenAI is initially making Dots available to subscribers on its Pro, Business Premium, and Enterprise plans in select markets. Each of these subscriptions will include one Dot at no additional cost, with an initial allowance for extended work during the first month. The company stated it plans to allow users to purchase additional Dots and more processing capacity in the future.

For larger organizations, OpenAI is piloting specialized agents, or "specialist Dots," which operate with their own distinct identity and credentials to handle specific corporate functions. The company noted it is working with Microsoft to integrate these specialist agents into a product called Agent 365 as part of focused enterprise trials.

Security and Data Controls

To address security concerns, Dots operate with safeguards such as restricted read-only access for background tasks and monitoring systems that can pause operations if potential safety issues arise. Users retain control over which applications the agents can access and can establish custom rules for when a Dot should act independently versus when it must seek approval.

OpenAI also reiterated its data privacy policies, stating that content from Business, Enterprise, and Education workspaces is not used to train its models by default. Users on personal plans will have the option to control whether their conversations with Dots are used for model improvement.