⚡ Quick Answer
OpenAI Dots are personalized, always-on AI agents that pursue user-defined goals and complete multi-step tasks across connected apps, browsers, and cloud computers. They can work in the background, provide progress updates, ask questions, and request approval before sensitive actions.
What OpenAI Dots Are and Why They Matter
Powered by OpenAI’s GPT-6 Astra model, Dots are intended to operate across connected software rather than simply answer questions. Users can name and customize a primary Dot, provide ongoing instructions, and allow it to learn preferences over time. The agent can then monitor information, make progress, ask questions when needed, and return completed work for review. That approach puts Dots in a different category from a single-purpose assistant. Instead of asking an AI to summarize one document or write one message, a user might assign it a continuing objective—such as tracking a project, reviewing customer feedback, or helping manage software updates.
How Dots Handle Multi-Step Work in the Background
Potential examples include monitoring a project for changes, processing incoming feedback, building and testing an update, or preparing a report from information spread across several tools. The practical advantage is continuity: Dots can continue working while the user focuses elsewhere, rather than losing context when a conversation ends. The model still matters, but access and execution may matter more. A capable agent with the right permissions could complete a workflow; an agent limited to reading a chat window could only describe what a person should do next. Those approval boundaries will be central to whether always-on agents feel useful or risky. Users need enough automation to avoid micromanaging every step, but enough visibility to catch errors before they affect accounts, data, customers, or production systems.
Availability, Pricing, and Early Limitations
Dots began rolling out through ChatGPT to paid users in eligible markets. Reporting identifies OpenAI’s $100-per-month Pro tier as one access route, while Business Premium users are also reported to be included in the rollout. Availability and features may vary by plan and market. The first major limitation is scale: users initially control one Dot. OpenAI plans to support multiple Dots working together, but that capability is not yet the starting point. A single agent may be sufficient for personal productivity, but businesses with separate research, engineering, support, and administrative workflows may quickly want specialized agents. Access is also part of the early product story. Dots are available in ChatGPT, with reporting describing access through workplace and messaging platforms such as Slack, Microsoft Teams, and iMessage, alongside possible phone-based interaction. The broader the access, the more important consistent permissions and audit trails become.
What Dots Could Mean for AI Assistants
But continuous operation does not automatically equal productivity. Agents must know when to proceed, when to pause, and when to escalate. Poor assumptions, stale information, or an incorrect action repeated across connected apps could create more work than the original task. For Dots, the early tests will be practical: how often they need human intervention, how clearly they explain progress, how narrowly users can limit permissions, and whether one Dot can manage complex workloads. If OpenAI gets those controls right, Dots could make AI feel less like a chatbot and more like a dependable digital coworker. Want to see how persistent AI could change everyday work? Explore our latest coverage of ChatGPT agents, connected-app automation, and AI safety controls.
Step-by-Step Guide
- 1
Define a persistent objective
Give the Dot a clear ongoing goal, expected outcome, priorities, and boundaries instead of issuing disconnected one-time prompts.
- 2
Connect the required tools
Authorize only the applications, browser sessions, files, and cloud environments the agent needs to complete the assigned workflow.
- 3
Set approval boundaries
Require explicit confirmation for sensitive actions such as installing software, changing passwords, sending messages, or modifying production systems.
- 4
Monitor progress and answer questions
Review status updates, supply missing information when the Dot pauses, and correct assumptions before the task continues.
- 5
Review the completed work
Check outputs, activity history, permissions, and downstream changes before accepting the result or allowing it to inform future work.
Key Statistics
Frequently Asked Questions
Key Takeaways
- ✓OpenAI Dots are designed to work continuously toward goals rather than respond only to individual prompts.
- ✓Dots can reportedly use connected applications, browsers, and cloud computers to complete multi-step workflows.
- ✓The agents can provide updates, ask follow-up questions, and seek approval before sensitive actions.
- ✓Initial access is reportedly limited to eligible paid users and one Dot per user, with availability varying by market and plan.
- ✓Dots’ long-term value will depend on reliable execution, narrow permissions, clear audit trails, and effective human oversight.