Oct 01, 2026 DISPATCH // HARDWARE, CODE & PLATFORMS

OpenAI unveils Dots: AI agent that runs your workflow on its own

OpenAI has launched Dots, an AI agent that takes over tasks and keeps users in the loop. This marks a clear move away from chatbots and toward real AI workflow automation.
OpenAI unveils Dots: AI agent that runs your workflow on its own Nerds Magazine © nerdsmagazine.com
OpenAI unveils Dots: AI agent that runs your workflow on its own © nerdsmagazine.com

Dots is already running inside OpenAI. One Dots agent now handles software bug reports from Slack. It coordinates backend fixes, iOS updates, testing, and reviews. No one needs to guide it step by step. Another Dots agent prepares for internal meetings. It follows up with presenters and spots missing slides. People at OpenAI say it saves hours.

OpenAI revealed Dots at Devday 2026 in San Francisco. This is not just another chatbot. Dots is built on the GPT-6 Astra model. It works nonstop. It tracks progress, adapts to changes, and reaches out to users when a decision is needed. Dots remembers past chats, user preferences, and ongoing projects. It can juggle several workflows at once. No constant supervision needed.

OpenAI states that Dots can connect with over 4,000 third-party applications through ChatGPT, enabling broad integration across business tools and platforms.

Dots links up with connected apps, Codex, and ChatGPT Work. It uses its own cloud computer for research, data analysis, documentation, and software development. Users can talk to Dots by text or voice. It works on web, mobile, desktop, Slack, and Microsoft Teams. If a task starts in one place and new info comes in somewhere else, Dots keeps the full context. Nothing gets lost.

OpenAI set strict limits. Users decide what Dots can do on its own, what needs a green light, and what is off-limits. Sensitive actions like changing passwords, deleting data, or installing software are blocked or need direct approval. At launch, Dots is for ChatGPT professional users in the European Economic Area, Switzerland, and the UK. Each user gets one Dot for now. OpenAI plans to let teams use several Dots together later.

According to OpenAI's official announcement, Dots are always-on agents available 24/7 in ChatGPT across web, mobile, and desktop for Pro, Business Premium, and Enterprise users in permitted markets. The rollout also includes integration with Slack and Microsoft Teams, with text messaging support planned for a later stage.

OpenAI

OpenAI also rolled out GPT-6.1 Sol. This new model is faster and cheaper than GPT-6 Astra. The API costs $2 per million input tokens and $10 per million output tokens. There is a 'cash input' option at $0.10, which is 95% less than the standard input price. Safety scores are up. AI attempts to break user rules dropped from 64.4% to 23.5%. Unauthorized actions fell from 17.4% to 4.3%. These numbers come from test environments, not real-world use.

ChatGPT pricing has changed too. The new Pro plan costs $500 per month. It unlocks 'UltraFast' speeds and higher answer rates. The old $200 Pro plan now has lower usage limits. ChatGPT Work and Codex access drops from 20 to 10 times the Plus plan. GPT-6 Pro message limits are cut to 100 per week. Current subscribers get one-time credits and keep their limits during the switch.

OpenAI's reach is huge. There are over 1.2 billion weekly ChatGPT users. ChatGPT Work and Codex see 35 million weekly users. More than 2.5 million companies use OpenAI products. The company is not just tweaking features. It is changing how AI works. Dots is a turning point. AI is no longer just waiting for commands. Now it can run real tasks, as long as users set the rules. The future of AI is clear. It is about real-world work and user control.

Topics:
Artificial Intelligence AI Tools #ChatGPT #AI Agents
Evan Solberg Technology publisher and editor-in-chief Nerds Magazine
Editor-in-Chief

Evan Solberg

Evan Solberg is the Founder, Owner, Publisher, and Editor-in-Chief of NerdsMagazine, where he covers consumer technology, software, artificial intelligence, privacy, and digital products. His editorial approach focuses on what technology actually does for readers, what it costs, where it falls short, and which claims deserve closer scrutiny.