- OpenAI Dots AI helper: Always-on agents launched at DevDay, powered by GPT-6 Astra
- Dedicated cloud computer: Every Dot runs its own browser and machine, independent of your device
- 4,000+ app connections: Link Slack, email, calendars, and workspace tools
- Scheduled tasks: Set reminders, recurring checks, and proactive follow-ups
- Permissions first: Control autonomy, connected apps, and workspace access before assigning work
What the OpenAI Dots AI Helper Is
The OpenAI Dots AI helper is an always-on AI agent product that OpenAI announced on September 29, 2026, during DevDay. Unlike a normal ChatGPT conversation that ends when you close the tab, a Dot is a persistent assistant driven by GPT-6 Astra, OpenAI's flagship model for complex reasoning, computer use, and research tasks. Each Dot gets its own cloud computer and browser, connects to external tools, and keeps working on long-running goals across channels like ChatGPT, email, and Slack.
The core difference is continuity. A regular chat answers one prompt. A Dot holds a standing goal, schedules its own work sessions, and follows up with you when results are ready.
A Dot is not a longer chat. It is an agent with a dedicated environment, connected apps, and a task list that persists between your conversations.
| Feature | Regular ChatGPT Chat | OpenAI Dots AI Helper |
|---|---|---|
| Lifespan | Ends with the conversation | Always-on, persists across sessions |
| Execution | Runs while you wait | Works in its own cloud computer |
| Apps | Limited to chat context | 4,000+ app connections |
| Follow-ups | You must re-prompt | Proactive updates via ChatGPT, email, Slack |
| Underlying model | Varies by plan | GPT-6 Astra |
OpenAI Dots Setup: Step-by-Step
Dots are created and managed from ChatGPT on desktop web or the supported desktop app. A well-configured Dot starts with a focused purpose, clear working rules, and the right level of autonomy.
Create Your Dot
Open the Dots experience in ChatGPT and start a new Dot. Give it a short, descriptive name tied to the project or recurring job it will own.
Define the Goal
Describe the ongoing outcome you want, not a single prompt. A focused goal gives the Dot stable direction across multiple work sessions.
Set Working Instructions
Specify what the Dot should prioritize, boundaries it must follow, and how independently it should operate before asking for approval.
Attach Context and Files
Add project documents and reference material so the Dot has the background it needs from the first task onward.
Assign the First Task
Give it something concrete connected to the main goal. The Dot uses its workspace and connected tools to work through it.
Review the Activity Dashboard
Open the Dot's activity view to follow current tasks, inspect completed actions, and adjust configuration as it works.
Dots access is plan-dependent and rollout is gradual. Desktop web and desktop app are the primary surfaces; mobile access varies by account and rollout stage, and advanced configuration is centered on desktop.
| Setup Element | Purpose | Example |
|---|---|---|
| Name | Identifies the Dot's responsibility | "Market Research Dot" |
| Goal | Ongoing outcome to work toward | "Track competitor pricing weekly" |
| Instructions | Priorities and boundaries | "Only summarize changes above 5%" |
| Autonomy | How independently it acts | Ask approval before sending emails |
| Files | Project context | Reports, spreadsheets, style guides |
Cloud Computer and Connected Apps
The cloud computer is what separates the Dots AI helper from chat-based assistants. Each Dot operates its own remote environment with a browser, so multi-step web work continues even when you close your laptop.
Dedicated Cloud Environment
- Own remote computer and browser
- No dependency on your local device
- Built for long-running workflows
Independent Execution
- Keeps working while you do other things
- You can view the session anytime
- Step in manually when review is needed
Connected Apps
- 4,000+ app connections available
- Slack, email, files, workspace tools
- Each link follows your granted permissions
Cloud and local access are separate capabilities. The dedicated cloud computer handles remote agent execution; optional local-computer access lets a Dot interact with your own machine only if you explicitly allow it.
| Connection | Category | What the Dot Can Do |
|---|---|---|
| Slack | Team communication | Work with permitted channels and context |
| Communication | Handle approved messages and follow-ups | |
| Files | Knowledge and context | Use attached documents as working reference |
| Workspace apps | Productivity | Extend workflows with external tools |
| Channels | Notifications | Send task updates and results proactively |
Every connected app inherits the permissions granted by you or your workspace administrator. A Dot cannot exceed the access level of the account that authorized the connection.
Automations, Scheduled Tasks and Use Cases
Scheduled tasks turn a Dot into an ongoing assistant. Work can run at a set future time or repeat on a schedule, and results arrive without a new prompt.
Create a Scheduled Task
Tell the Dot what should happen and when — a one-time future action or a recurring schedule.
Run Recurring Checks
Use schedules for repeated work: reviewing updates, checking connected data, preparing weekly summaries.
Receive Proactive Updates
The Dot surfaces results through your configured channels once scheduled work completes.
Review Task Activity
The task management view shows active, scheduled, and completed work in one place.
The best OpenAI Dots use cases share one trait: work that benefits from persistence rather than a single answer.
| Use Case | How the Dot Helps |
|---|---|
| Ongoing research | Collects, organizes, and updates findings across sessions |
| Calendar review | Reviews upcoming events and prepares recurring summaries |
| Reminders and follow-ups | Surfaces important tasks at the right time |
| Recurring monitoring | Watches changing information and reports meaningful updates |
| Connected-app workflows | Chains actions across Slack, email, and other permitted apps |
| Multi-step agentic work | Handles research, browser tasks, and follow-up end to end |
Give each Dot one clear domain. A single research Dot plus a single scheduling Dot outperforms one overloaded Dot trying to do everything.
Permissions and Workspace Controls
Permissions define the boundaries a Dot operates within. Both individual users and workspace administrators can restrict what each Dot accesses.
| Control | Scope | What It Governs |
|---|---|---|
| Connected app permissions | User and workspace | Which apps and services the Dot may access |
| Cloud computer access | Dot capability | Whether it can run browser and multi-step tasks |
| Local computer access | Optional user capability | Separate control for your own device |
| Autonomy settings | User configuration | How independently the Dot acts between check-ins |
| Workspace availability | Administrator | Whether members can use Dots at all |
| Enterprise controls | Enterprise admin | Organization-level restrictions and governance |
Review connected-app permissions and autonomy settings together. A Dot that can act autonomously and send email on your behalf should have explicit approval rules for sensitive actions.
Before Assigning Long-Running Work:
- Define a single clear goal per Dot
- Connect only the apps it needs
- Set approval rules for sensitive actions
- Confirm autonomy level matches task risk
- Check the activity dashboard regularly
- Revoke access for retired workflows
FAQ
Q: What is the OpenAI Dots AI helper?
OpenAI Dots are always-on AI agents launched on September 29, 2026, powered by GPT-6 Astra. Each Dot has its own cloud computer and browser, connects to apps, and keeps working on long-running tasks between conversations via ChatGPT, email, and Slack.
Q: Which platforms support creating Dots?
Dots are created and configured through ChatGPT on desktop web or the supported desktop app. Mobile access depends on account, platform, and rollout stage, with advanced configuration centered on desktop.
Q: How is a Dot different from a normal ChatGPT conversation?
A normal chat handles one prompt and ends. A Dot holds a persistent goal, runs tasks on a schedule in its own cloud computer, uses connected apps, and sends proactive updates instead of waiting for you to re-prompt.
Q: Can administrators restrict Dots in a workspace?
Yes. Business and Enterprise workspaces can manage whether members can use Dots, control connected tools, apply autonomy limits, and enforce organization-level governance through admin settings.