OpenAI announced dots at their DevDay event. Dots are persistent AI agents powered by GPT-6 Astra. Each dot runs on a separate cloud computer with a browser. They work across more than 4,000 apps through ChatGPT plugins.
Dots turn ChatGPT into a standing agent. You give it a goal, a name, and boundaries. It then works around the clock to achieve that goal. It learns from your feedback to improve over time.
Each user starts with one primary dot. Dots can be used for research, analysis, documents, and software tasks. They can also start tasks in Codex and ChatGPT Work. You can message or call your dot on web, desktop, or mobile. Support for Slack and Teams is available, with texting coming soon.
The underlying model is GPT-6 Astra. In latency tests, it scored 72.6% on OSWorld 2.0, completing tasks in about 40 minutes. This is faster than GPT-5.6 Sol, which scored 65.7% and took about 75 minutes.
OpenAI includes safety features like custom rules, auto-review, and safety monitors. Dots can pause or stop if safety concerns arise. They sign in with saved passwords without exposing sensitive data.
Specialist dots are also in preview. They have their own identity and access to company systems. Enterprise pilots are starting, with support for Microsoft Agent 365 planned.
Dots are available now for Pro and Business Premium plans at no extra cost. Conversations with a dot do not count toward usage limits. OpenAI also added a paid tier for $500. These tools aim to improve productivity while maintaining safety.
Why it matters
This development offers continuous, goal-oriented AI agents that can operate safely and efficiently across many applications.
What to do
Evaluate if persistent agents fit your workflow. Consider testing dots in your enterprise pilot programs.