OpenAI unveiled Dots, a new AI assistant platform designed to function as always-on agents that operate across multiple applications while adapting to user preferences over time. The announcement was made during the company's DevDay conference on Tuesday. Dots are powered by OpenAI's GPT-6 Astra model and run on a dedicated cloud computing environment that provides access to a web browser and more than 4,000 integrated applications.

Users can engage with Dots through a text-based interface resembling a messaging thread, initiate voice conversations via ChatGPT on web, desktop, or mobile platforms, and connect directly to workplace tools such as Microsoft Teams and Slack. In professional settings, Dots can inherit context from prior interactions across different apps and devices. Future capabilities will include SMS-based communication.

Currently, each user can create a single Dot, though OpenAI has indicated plans to expand functionality to support multiple agents operating simultaneously. Once assigned a task, a Dot provides progress updates and prompts for input when necessary. For example, a software developer could task their Dot with analyzing customer feedback to generate and test software updates, while simultaneously producing a demonstration video of the changes. Similarly, a content creator might direct their Dot to review interview transcripts, identify key moments for editing, draft show notes, and compose social media content.

Users can monitor a Dot's activities through its cloud-based computing interface. The system is designed to learn from interactions and refine its performance over time, potentially suggesting tasks before users explicitly request them. Collaborative workflows with Dots are supported within ChatGPT Space, a new shared environment for team-based interaction.

The introduction of Dots follows Meta's recent release of Muse, an AI assistant capable of web navigation, transaction processing, document generation, and goal tracking. While Muse has gained significant traction, it has also faced scrutiny over safety concerns, including an incident where it reportedly shared a user's address with a buyer on Facebook Marketplace without authorization.

OpenAI has emphasized that Dots operate under predefined behavioral constraints and include an automated verification system to ensure compliance with custom rules and safety protocols. This mechanism determines when a Dot should execute an action independently versus when it should defer to user input, such as when changing account credentials. Initial access to Dots is being rolled out to subscribers of ChatGPT Pro, Business Premium, and Enterprise plans. Enterprises are also being invited to develop specialized Dots tailored to specific organizational functions.

OpenAI has noted that interactions with Dots will not consume standard ChatGPT usage quotas. The company has indicated plans to eventually support additional Dots and increase their operational capacity through performance enhancements or expanded workload allocations per user.