OpenAI has introduced AI dots, a new type of always-on agent built to handle a wide range of work. The company says they learn what matters to each user and work on their behalf. The goal is to take important tasks off people’s plates and give them time and attention back.

What Are Dots?

Dots are powered by GPT-6 Astra. Each one has its own cloud computer. They learn from feedback over time and can work toward goals 24/7.

A plugin ecosystem lets dots connect to more than 4,000 apps. This gives them tools to help in many areas.

Users can reach a dot through ChatGPT, Slack or Teams. They can ask questions, explore ideas and give feedback. They can also start a voice call to talk something through.

Over time, a dot learns a user’s preferences and how they think. It also learns what good work looks like to them. As a result, it can deliver work the way the user would do it, sometimes before being asked.

How OpenAI Uses Dots Internally

OpenAI says it is seeing what work looks like when intelligence is always on. For example, a bug appears in Slack and dots start investigating. A new design arrives and dots turn it into a working app. During a planning cycle, dots help keep everyone in sync toward the deadline.

An early outside tester also shared an example. His dot noticed he had forgotten to invoice a publication. It prepared the invoice and sent it after his approval.

Rollout and Future Plans

Users can start with a primary dot, give it a name and make it their own. OpenAI says it envisions teams of dots working together in the future.

The rollout begins across Pro, Business Premium and Enterprise plans in eligible markets. OpenAI plans to expand to more users soon.

The company is also sharing a preview of specialist dots. These have their own identity for access management. They use IT-provisioned hardware. They also support deep integrations with a company’s systems of record, so they can take on well-defined responsibilities inside organizations.

AI Dots Can Work on Their Own

Dots can do nearly anything using their own cloud computer, their own browser and the apps a user has connected. Users can open a dot’s computer at any time to inspect its work.

Dots can also connect to other devices. Users can give a dot permission to use their laptop. There, it can work directly alongside them.

A dot can take on a project and run with it. It can do this while handling several others. Users can keep giving it new tasks and ideas. They do not need to juggle separate threads or direct every step.

Five Ways Dots Extend the User

A dot that knows a user’s goals and standards lets them take on more without directing every step. OpenAI describes five examples.

Turn feedback into tested fixes. A developer’s dot watches customer feedback for recurring requests. It scopes smaller improvements and bugfixes, then builds and tests them. It brings back complete PRs with attached videos showing the changes. Meanwhile, the developer can keep building the next major feature.

Revise a launch as scope changes. A product launch lead’s dot learns the audience, positioning and creative standards. When scope changes, it works out what that means for the story. It then revises launch materials and prepares drafts of asset and doc changes. The lead returns to a plan ready for review.

Update analysis with new evidence. A scientist’s dot learns the research question and how they evaluate evidence. As new data arrives, it reruns analyses and investigates unexpected results. It also updates figures for the paper. When findings change, it revises the supporting explanation and flags what needs review.

Update a proposal as requirements shift. A sales lead is guiding an enterprise deal through technical review. Their dot checks customer requirements and account history against product docs. It finds what still needs testing and builds a proof of concept for a key integration. It also suggests a solutions engineer who has handled a similar deal. As requirements and test results change, it updates the proposal and test plan. It flags open concerns as well. The lead and engineer focus on customer conversations and approve commitments.

Automate content production work. A content creator’s dot learns their voice and what matters to their audience. When a new interview transcript arrives, it identifies moments for clips. It also prepares show notes and drafts social posts for approval. It carries edits across the materials and learns what resonates.

Where Users Can Reach Their Dot

Working with a dot is as simple as having a conversation. Users can message or call dots in ChatGPT on desktop, web and mobile. Dots can also message users with progress, questions or decisions that need them.

Users can message a dot in Slack and Teams too. Texting is coming soon. A user can start a project in ChatGPT and share context with the working team in Slack. The dot then follows through. Dots carry context across every channel, so they have the full picture wherever they are reached.

Safety, Privacy and Control

According to OpenAI, dots build on ChatGPT’s protections with additional safety and privacy safeguards. Users can set boundaries, follow progress and stay involved in decisions that need them.

Built-in safeguards

Each dot works on its own cloud computer. A user’s computer and its contents stay separate unless the user chooses to connect it. When signing into supported websites, dots can use saved passwords without exposing them to the model.

When a user is not actively working with it, a dot looks for ways to help in the background. OpenAI calls this “proactive research.” It uses apps the user has already connected. The tools involved are restricted to read only. They cannot send messages, change app content or control the browser or computer.

Security safeguards also help defend against malicious instructions and monitor for potentially harmful behavior. If the monitoring system detects a safety concern, it can pause or stop the dot’s work.

Access and permissions

Users choose which apps dots can access. They can manage permissions through existing ChatGPT app controls.

Dots start with built-in rules for when to act independently and when to ask for approval. Custom Rules let users allow specific actions, require approval or block them. Built-in safety requirements always apply. Users can follow progress, including background work, in Activity View. They can redirect dots as needed.

On data use, OpenAI says it does not use content from ChatGPT Business, Enterprise or Edu workspaces to improve its models by default. On personal ChatGPT plans, users can control whether dots’ conversations and work are used to improve models. The company does not train directly on proactive research or a dot’s notes to itself. However, information from them may be used if it helps inform an eligible conversation or task, depending on settings.

Action review and approvals

Dots use [auto-review] to check actions that could affect a user’s accounts or share information. These checks compare actions against the user’s instructions, Custom Rules and safety requirements. The process helps decide what can proceed, what needs approval and what the user must do. Certain sensitive tasks, such as changing a password, always stay with the user.

More detail is available in the dots safety blog, the [Help Center] and the [system card]. OpenAI notes that dots can still make mistakes. Users should always review consequential work.

Specialist Dots for Organizations

A user’s dot works on their behalf. In contrast, specialist dots take on dedicated responsibilities within an organization. A company sets up each one with its own identity, credentials and access to the systems it needs. As it works, the dot adapts to new information and improves with feedback from the team.

OpenAI is building on lessons from early testing across several functions. These include procurement, invoice processing, email marketing, customer support and commercial contracting.

The company is starting with focused enterprise pilots. Its engineering teams will work directly with organizations. Together, they will define each dot’s responsibilities, the tools it can use and how people review and approve its work. Organizations with work they would like a specialist dot to take on can contact OpenAI.

Microsoft Agent 365 integration

OpenAI is also working with Microsoft to integrate specialist dots with governance and security controls in Agent 365. The goal is to let businesses manage dots through Microsoft tools they already use.

Availability and Pricing

Dots are rolling out today in ChatGPT to Pro and Business Premium users in eligible markets. Enterprise users, including Edu and Healthcare, can try the beta when their workspace admin enables it.

The first dot is included in a Pro or Business Premium plan at no extra cost. It is available 24/7 to talk, help users think and stay on top of what matters to them. The plan also includes an allowance for deeper work. Limits are extended for the first month after launch.

In the future, users will be able to add more dots. They will also be able to scale each dot’s output. This can be done by increasing its speed or the total amount of work it can take on per month.

Conversations with a dot do not count toward ChatGPT usage limits. However, tasks a dot starts or manages in Codex or ChatGPT Work count toward usage limits as usual.

How to Get Started

Users can create their first dot in the ChatGPT desktop app or a desktop browser. Next, they connect their apps and let the dot introduce itself. After the initial setup, they can message it in the ChatGPT mobile app.

OpenAI says it will keep improving AI dots as it learns what makes them useful every day. It plans to share what it learns where it can and use those lessons to guide what it builds next.

Related post

View all