OpenAI’s Always-On AI Agents Redefine Automation
Meet Dots: OpenAI’s latest AI agents that work independently across 4,000+ apps. Available now for ChatGPT Pro and Business Premium subscribers in eligible markets.
AI agents for work
OpenAI Launches Dots: Your AI Agent That Never Stops Working
Imagine having a dedicated assistant that operates around the clock, handling routine tasks across thousands of applications without you needing to lift a finger. That’s the promise of Dots, OpenAI’s newest offering announced in late September 2026 at the company’s annual developer event in San Francisco. These always-on AI agents represent a fundamental shift in how people can delegate work to artificial intelligence.
Built on GPT-6 Astra, OpenAI’s latest model, Dots function as autonomous helpers that can manage projects, communicate updates, and even make decisions within predefined boundaries. The rollout has begun for ChatGPT Pro and Business Premium subscribers in eligible markets, with the first dot included at no additional cost. Enterprise users, including those on Edu and Healthcare plans, can access a beta version once their workspace administrator enables it.
What Dots Actually Do and How They Work
Each dot operates with its own cloud computer and browser, connecting to applications through OpenAI’s plugin ecosystem. The company describes these agents as capable of handling “nearly anything” within their connected apps. Users retain complete visibility—you can open a dot’s computer environment at any time to review its work, and you can even grant it access to your own laptop if needed.
The setup is straightforward. You create one primary dot, give it a name, and define its responsibilities. Over time, the agent learns your preferences and work patterns, anticipating tasks before you explicitly request them. A single dot can juggle multiple projects simultaneously, sending you updates and flagging decisions that require your approval.
Real-world examples illustrate the practical applications. A software developer’s dot could monitor customer feedback, construct and test code fixes, then prepare them for review complete with video demonstrations of the changes. A podcaster’s dot might extract clips from interview transcripts, write show notes, and draft social media announcements. One early tester reported that their dot noticed an unpaid invoice, created the invoice automatically, and sent it for approval—catching a task the user had genuinely forgotten.
Access and Communication Across Multiple Platforms
Dots integrate seamlessly with your existing communication workflows. You can message or call your agent directly through ChatGPT on desktop, web, and mobile. Integration with Slack and Microsoft Teams means your dot remains accessible from wherever you work. Text messaging support is coming soon. The agent maintains full context across all these channels, so switching platforms doesn’t lose information or break workflow continuity.
There’s an important setup requirement: you must configure your dot using either the ChatGPT desktop application or a desktop web browser initially. Once established, you can reach it from mobile apps without re-configuring. One useful note for heavy ChatGPT users: conversations with your dot don’t count toward your ChatGPT usage limits, though tasks launched in Codex or ChatGPT Work do.
OpenAI hints at future expansion too. The company stated that eventually you’ll be able to add multiple dots and scale each one’s output by increasing its speed or the total workload it handles monthly. This points toward users potentially deploying specialized agents for different functions.
Smart Safety Controls Keep Your Work Protected
OpenAI built several safeguards directly into Dots to prevent misuse or unintended actions. Each agent comes with built-in rules determining when it acts independently versus when it must ask for your approval. The Custom Rules feature lets you explicitly allow, require approval for, or block specific actions entirely.
An auto-review system scrutinizes any action that could affect your accounts or share information, checking it against your instructions, your custom rules, and OpenAI’s safety requirements. Sensitive operations—like changing passwords—always remain under your control and never happen without explicit approval. When idle, Dots perform what OpenAI calls “proactive research,” which limits itself to read-only tools that cannot send messages, alter app content, or control browsers or computers.
The company operates a monitoring system capable of pausing or stopping a dot if safety concerns emerge. Additionally, Dots can sign into supported websites using saved passwords without the model ever seeing those passwords directly. That said, OpenAI emphasizes that Dots can still make mistakes, so you should always review consequential work before it takes effect.
Regarding data privacy, OpenAI does not use content from Business, Enterprise, or Edu workspace interactions to train its models by default. Users on personal plans have the choice of whether their dot’s work contributes to model training.
Enterprise Specialist Dots for Organizations
Beyond consumer-focused Dots, OpenAI is previewing specialist dots designed for organizational use. These agents assume defined roles within companies, each with its own identity for access management and unique credentials. They can integrate deeply with a company’s internal systems of record, handling specialized functions across different departments.
OpenAI has tested specialist dots internally across procurement, invoice processing, email marketing, customer support, and commercial contracting. The company is beginning rollout through enterprise pilot programs. Notably, OpenAI is partnering with Microsoft to bring specialist dots under the governance and security controls of Microsoft’s Agent 365 platform, a significant step for organizations managing AI adoption across their infrastructure.
How Dots Compare to Competing AI Agents
Dots face competition from Muse, the AI agent Meta released two weeks prior. Muse can browse the web, make purchases, create documents, and track goals, positioning itself mainly for consumers. The Muse app launched in the US and has exceeded 3 million downloads, though the service has already faced privacy concerns—one user reported Muse shared their address to a Facebook Marketplace buyer without authorization.
OpenAI’s agents target businesses more directly, though Meta did announce a small business offering alongside Dots’ launch. This suggests both companies see organizational and freelance markets as crucial growth areas. The distinction matters: Dots’ integration with ChatGPT Pro and Business Premium, plus specialist dots with enterprise controls, position OpenAI’s solution differently from Meta’s consumer-first approach.
Key Questions About Dots You Should Know
Are Dots available to all ChatGPT users? No. Dots rollout to ChatGPT Pro and Business Premium subscribers in eligible markets. Enterprise, Edu, and Healthcare users can access a beta version through their workspace administrator. Availability may be limited by geography and account tier.
Is there an extra cost for the first Dot? The first dot comes included with both ChatGPT Pro and Business Premium plans at no additional charge, according to OpenAI’s announcement.
Will Dots work on my phone? Yes, but with a caveat. You must set up your dot using the ChatGPT desktop app or desktop browser first. After that initial setup, you can access it from mobile apps, as well as through Slack and Microsoft Teams.
What to Try Next
If you’re a ChatGPT Pro or Business Premium subscriber in an eligible market, check whether your account has access to Dots yet. Start by setting it up through the ChatGPT desktop application and define a specific repetitive workflow—invoicing, content research, or customer feedback tracking work well for initial testing. Remember to review the Custom Rules feature and set explicit boundaries for actions you want to approve manually.



