OpenAI's Dots: When AI Stopped Waiting for Instructions and Started Working 24/7

For years, using AI meant typing a prompt and waiting for a response. You asked, it answered. You stopped, it stopped. That era just ended.

OpenAI has launched Dots — always-on autonomous agents powered by GPT-6 Astra that don't wait for instructions. They have their own cloud computer, their own browser, connections to over 4,000 apps, and the ability to work toward your goals around the clock. The announcement, which hit the front page of Hacker News with over 600 points and nearly 500 comments, represents something genuinely new in the AI landscape.

This isn't another chatbot upgrade. It's a fundamental shift in how we interact with AI — from conversation to delegation.

What Are Dots, Exactly?

A Dot is a persistent AI agent that runs on its own cloud computer. It learns your preferences over time, connects to the apps you already use, and can handle complex multi-step work without you directing every move. Think of it less as a tool and more as a digital employee who never sleeps.

The core capabilities include:

• A dedicated cloud computer where the Dot runs its own browser and executes tasks independently
• Integration with 4,000+ apps through OpenAI's plugin ecosystem
• Cross-channel communication — you can reach your Dot through ChatGPT, Slack, or Microsoft Teams, with texting coming soon
• Voice calls for when you need to talk something out rather than type it
• Context that carries across every channel, so your Dot never loses the thread

The 'Always Working' Problem

The most striking feature of Dots is what OpenAI calls 'proactive research.' When you're not actively talking to your Dot, it doesn't just sit idle. It looks for ways to help in the background — monitoring connected apps, looking for things you might have missed, and preparing work for your review.

OpenAI shared a compelling example: an early tester's Dot noticed he had forgotten to invoice a publication. It prepared the invoice and sent it after getting his approval. At OpenAI internally, a bug appears in Slack and Dots immediately start investigating. A new design arrives and Dots turn it into a working app while the team focuses on customer feedback.

This is the promise that every 'AI assistant' has hinted at but never delivered: an AI that doesn't just respond to you, but anticipates you.

Safety: The Guardrails That Matter

An always-on AI that can send messages, interact with apps, and work autonomously is also an always-on AI that can make mistakes at scale. OpenAI knows this. Dots ship with several layers of protection:

• Each Dot runs on its own cloud computer, separate from your local machine — your computer and files stay private unless you explicitly connect them
• Saved passwords can be used to sign into supported websites without exposing them to the AI model itself
• Proactive research is restricted to read-only tools — background work can't send messages, change app content, or control your browser
• Built-in safety monitoring watches for malicious instructions and can pause or stop a Dot's work if it detects something concerning
• Custom Rules let you allow specific actions, require approval, or block them entirely — certain sensitive tasks like changing passwords always stay with you

There's also an auto-review system that checks actions against your instructions and safety requirements before they execute. An Activity View lets you follow everything your Dot is doing, including background work, and redirect it if needed.

Specialist Dots: AI as Organizational Infrastructure

Beyond personal Dots, OpenAI is previewing 'specialist Dots' — agents with their own identity, credentials, and system access that take on dedicated responsibilities within organizations. Think of them as AI employees with specific job descriptions.

OpenAI has been testing specialist Dots internally across procurement, invoice processing, email marketing, customer support, and commercial contracting. Each one is set up with defined responsibilities, the tools it can use, and clear approval workflows. Companies work directly with OpenAI's engineering teams to define these boundaries.

This is where things get interesting for businesses. If a specialist Dot can handle invoice processing end-to-end — reviewing submissions, checking against contracts, routing for approval — that's not an efficiency improvement. That's a role elimination. And it's a role elimination that comes with full audit trails, consistent application of rules, and 24/7 availability.

The Competitive Landscape Just Shifted

Dots put OpenAI in direct competition with a growing field of autonomous agent platforms. Anthropic has been researching multi-agent systems. Google launched AX for agent infrastructure. Startups like Pion and Andon Labs have been letting AI run real companies. But Dots are different in one crucial way: they ship with the full force of OpenAI's ecosystem behind them.

When you combine GPT-6 Astra's frontier intelligence, 4,000+ app integrations, Slack and Teams presence, and the existing ChatGPT user base of hundreds of millions, you get something no startup can match: an autonomous agent platform that's immediately useful to a massive audience on day one.

What This Means for You

Dots are rolling out now to Pro, Business Premium, and Enterprise plans in eligible markets, with broader access planned. If you're in one of these tiers, you can start with your primary Dot, give it a name, and begin teaching it your preferences.

Here's what to think about:

• Start small. Give your Dot a contained task with clear success criteria before expanding its autonomy
• Set Custom Rules early. Decide what your Dot can do without asking, what needs approval, and what's off-limits
• Review the Activity View regularly, especially in the first weeks. Your Dot is learning, and so are you
• Think about which of your workflows involve repetitive multi-step work across multiple apps — these are prime candidates for Dot delegation

The Bigger Picture

Dots represent a shift from AI as a tool you use to AI as a presence that works alongside you. The distinction matters. A tool waits for you. A presence anticipates you. A tool executes commands. A presence pursues goals.

The question isn't whether Dots will be useful — the early examples make that clear. The question is whether we're ready for what happens when millions of people have an always-on AI working on their behalf. The privacy implications, the security surface area, the economic disruption, the shift in what 'work' even means — these are conversations that need to happen now, not after Dots are everywhere.

OpenAI has taken a thoughtful approach to safety, with layered protections, read-only background work, and approval workflows. But no safeguard is perfect, and an agent that works 24/7 has 24/7 worth of opportunities to get something wrong.

Still, this is the most ambitious agent product launched by a major AI company to date. And it's available today. The era of prompt-and-wait is over. The era of always-on AI has begun.