OpenAI Unveils Dots, New Always-On AI Agents for Complex Tasks

OpenAI has introduced Dots, a new line of always-on AI agents designed to handle complex tasks by accessing user files, apps, and context. Unveiled by CEO Sam Altman during the DevDay keynote, these agents operate similarly to a chief of staff or a high-level engineer. Unlike basic virtual assistants, Dots can manage multi-step workflows like monitoring bug reports, drafting code fixes, and managing project timelines without constant supervision.

Capabilities

These intelligent bots can learn work habits over time by accessing personal files and communicating with platforms such as Slack, Microsoft Teams, calendars, and other development environments. They connect with over thousands of apps using existing connectors and feature their own cloud-based computers and browsers. Users can interact with Dots using voice, text, or phone calls around the clock, and CEO Sam Altman demonstrated their use in application migration, product launches, and administrative delegation. Powered by OpenAI's Astra model, the agents work within defined safety boundaries set by the user.

Additional Launches

Alongside Dots, the company showcased Spaces, a collaborative platform that lets users and their Dots work together on documents, research, and data visualization. OpenAI also introduced GPT-6.1 Sol, a lower-priced model optimized for daily work, and UltraFast, a speed-boosted model for APIs, ChatGPT, and Codex. Additionally, Codex is now fully cloud-based, and the company launched the Marketplace for developers to distribute products and enterprise customers to make purchases.

Context

AI agents represent a shift from software that simply responds to prompts to tools that can independently execute multi-step workflows across different applications. Companies are increasingly building these capabilities into their subscription plans to make advanced automation features accessible to everyday users and businesses alike.