Short answer: Dots are OpenAI’s proactive, always-on AI agents, launched on 29 September 2026. Instead of waiting for a prompt, a dot works toward goals you give it, on its own cloud computer and browser, using the apps you connect. It sends progress updates, asks questions when it needs a decision and can be reached through ChatGPT, Slack and Microsoft Teams. At launch, Dots are available to ChatGPT Pro and Business Premium users in eligible markets, with an Enterprise beta and focused pilots of “specialist dots” for organisations.
Dots are a clear signal of where AI is going: from assistants you talk to, towards agents that work alongside your teams. Here is what that means in practice, and how to prepare.
According to OpenAI’s announcement, Dots are powered by GPT-6 Astra, learn from feedback over time and can connect to more than 4,000 apps through plugins. A dot can work on several projects at once, for example turning customer feedback into tested fixes, updating a sales proposal, refreshing a research analysis or preparing launch materials.
Each user starts with a primary dot that they can name and personalise. OpenAI describes a future where teams of dots work together on a person’s behalf, and where organisations deploy specialist dots with defined responsibilities.
The most important part for businesses is the specialist dot. These agents receive their own identity, credentials and access to systems, so they can take on a defined role inside an organisation, such as monitoring support requests or keeping reports up to date. OpenAI is starting with focused enterprise pilots and is working with Microsoft to integrate specialist dots with Agent 365 governance and security controls.
This is where AI agents stop being a productivity tool for one person and start becoming part of the organisation’s operating model.
| Chat assistant | Personal always-on agent (e.g. Dots) | Custom business agent | |
|---|---|---|---|
| Starts work | When prompted | Proactively, toward goals | On events, schedules or requests |
| Runs where | Vendor cloud | Vendor cloud, own sandboxed computer | Your cloud, private cloud or on-premises |
| Access to systems | Limited | Through plugins and connected apps | Through your APIs and MCP servers |
| Control of data and models | Vendor | Vendor, with admin settings | You decide, including open-weight models |
| Best for | Ad-hoc questions | Individual knowledge work | Core business processes and integrations |
Many companies will use all three: personal agents for knowledge workers, and custom agents for processes that depend on internal systems, strict data rules or specific integrations.
We help companies adopt agentic AI safely, whether that means preparing systems and governance for vendor agents such as Dots, or building custom agents on Claude, open-weight models and MCP connectors. Our Agentic AI & Cloud team designs permissions, approvals and observability, and our OnOptic case study shows agents working in a real business with human approval for sensitive actions. For the bigger picture, read our article on AI readiness.
Dots are proactive, always-on AI agents from OpenAI, launched on 29 September 2026. They work toward user-defined goals on their own cloud computer, use connected apps and report back through ChatGPT, Slack and Microsoft Teams.
At launch, ChatGPT Pro and Business Premium users in eligible markets can use Dots. Enterprise, Edu and Healthcare workspaces can try a beta when an administrator enables it, and specialist dots are available through enterprise pilots.
A specialist dot is an agent with a defined responsibility inside an organisation, with its own identity, credentials and system access. OpenAI is integrating specialist dots with Microsoft Agent 365 governance and security controls.
OpenAI provides safeguards such as separate cloud computers, read-only background research, custom rules for actions and an activity view, and states that business content is not used for training by default. Companies should still review data processing terms, permissions and approval rules before connecting business systems.
Use vendor agents like Dots for individual knowledge work where they fit. Build custom agents when a process depends on internal systems, strict data residency, specific integrations or models you need to control, including open-weight models.