OpenAI’s New “Dots” Aim to Become Always-On AI Coworkers That Work for You 24/7

OpenAI has introduced dots, a new type of always-on AI agent designed to work continuously on a user’s behalf. Unlike a traditional chatbot that primarily responds when prompted, dots can take ownership of projects, use their own cloud computer and browser, connect with thousands of apps, learn from feedback, and proactively perform work in the background.

Powered by GPT-6 Astra, dots are designed to understand a user’s goals, preferences, and standards over time. They can operate through ChatGPT as well as workplace tools including Slack and Microsoft Teams, while keeping users involved when important decisions or approvals are required.

OpenAI is beginning the rollout with Pro, Business Premium, and Enterprise users in eligible markets, while also previewing specialized enterprise dots that can take responsibility for defined organizational functions.

Key Points

• Dots are designed to work independently. Each dot has its own cloud computer and browser and can use connected applications. Users can inspect what a dot is doing and can optionally allow it to work on their own laptop.
• They are powered by GPT-6 Astra and can work toward goals around the clock. OpenAI says its plugin ecosystem allows dots to connect with more than 4,000 apps.
• Dots learn how users work. Through ongoing conversations and feedback, a dot can learn preferences, standards, goals, and what the user considers good work.
• They can manage substantial workflows. Examples in the article include investigating bugs, turning designs into working apps, coordinating planning cycles, creating tested software fixes, updating scientific analyses, preparing sales proposals, and producing content.
• Dots can be proactive. When users aren’t actively working with them, dots can conduct what OpenAI calls “proactive research” using connected apps with read-only tools.
• Users retain approval controls. Custom Rules can determine which actions a dot may perform independently, which require approval, and which are blocked. Sensitive tasks such as changing passwords remain with the user.
• OpenAI is developing specialist dots for companies. These agents can have their own identities, credentials, system access, and dedicated organizational responsibilities. Initial testing has included procurement, invoices, email marketing, customer support, and commercial contracting.
• Microsoft integration is planned. OpenAI says it is working with Microsoft to integrate specialist dots with Agent 365’s enterprise governance and security controls.
• The first dot is included at no additional cost for Pro and Business Premium users, although deeper work comes with an allowance and tasks delegated into Codex or ChatGPT Work continue to count toward their normal usage limits.

Key Quotes

“Dots are frontier intelligence that have your back.”

That line captures OpenAI’s positioning: dots aren’t presented simply as another conversational interface but as intelligence that can continuously work toward a user’s goals.

“The magic of dots is when they bring you work done the way you would do it, sometimes before you even think to ask.”

This highlights one of the biggest differences from conventional AI assistants: OpenAI wants dots to become sufficiently familiar with users that they can anticipate useful work.

“Your dot can take a project and run with it.”

Rather than forcing users to manage every individual AI conversation or step, dots are designed to maintain responsibility for projects while simultaneously handling other work.

“Dots can still make mistakes, so always review consequential work.”

Despite the emphasis on autonomy, OpenAI explicitly cautions users to review important work produced or performed by dots.

Implications

The article points toward a substantial change in how OpenAI sees people working with AI. Instead of users repeatedly opening ChatGPT, creating prompts, and managing separate conversations, dots are intended to maintain context, understand ongoing goals, and continue working even when the user isn’t actively interacting with them.

That could shift AI from being primarily an on-demand assistant to an ongoing collaborator. The examples OpenAI gives software development, product launches, scientific analysis, enterprise sales, and content production, show that dots are intended to handle multi-step professional workflows rather than isolated tasks.

The specialist-dot preview extends that idea further. Rather than every AI agent belonging to an individual, organizations could deploy dedicated agents with their own credentials and responsibilities. OpenAI’s pilots across procurement, invoicing, marketing, customer support, and contracting suggest dots could eventually function as persistent AI operators inside specific business processes.

At the same time, autonomy makes permissions and oversight increasingly important. OpenAI emphasizes cloud-computer isolation, read-only proactive research, Custom Rules, action review, approval requirements, and monitoring designed to pause potentially harmful activity. The article therefore presents greater AI autonomy alongside mechanisms intended to keep consequential decisions under human control.

Source: https://openai.com/index/introducing-dots/