OpenAI Launches Dots, Its Always-On Answer to Muse
OpenAI launched Dots at its annual Developer Day in San Francisco. The system arrived this week as OpenAI's answer to Muse. Reuters The Verge
OpenAI describes Dots as "remarkably capable, always-on agents" and as "proactive assistants that keep working across complex projects and everyday tasks." OpenAI Each dot is "an always-on agent in ChatGPT" that "keeps work moving across tools and projects." OpenAI Help OpenAI Codex
In functional terms, dots can "research, analyze data, and prepare documents." They can use a computer and pull information from connected applications to conduct research. They are designed to perform tasks for users over long periods. OpenAI Codex Bloomberg Yahoo Finance
The interaction model starts with character creation. Users make a Dot character, similar to making a Muse character. The positioning is explicitly business-first. The Verge
OpenAI markets Dots for business tasks including launching websites, rescheduling calls, and building slide decks for board meetings. That list matters. It is not general chat. It is cross-application office work, with artifacts and calendar state attached. The Verge
Distribution began immediately. OpenAI began rolling out Dots immediately upon announcement. Mashable Dots are available to Business Premium users across all supported ChatGPT regions. OpenAI Help They will be available on OpenAI's $100-per-month Pro plan, making $100 per month the minimum entry point. Mashable The Verge
OpenAI also unveiled a new $500 paid tier alongside Dots. CEO Sam Altman described Dots as a premium product. Bloomberg The company detailed the launch in an official announcement titled "Introducing dots," dated Oct 1, 2026. OpenAI OpenAI unveiled a potential Google Workspace competitor alongside Dots. Fortune
The broader context here is a shift from session-based assistance to persistent delegation. Chat ends when the window closes. A dot, by OpenAI's description, does not. It keeps work moving across tools and projects, which implies background execution, retained context, and permissions that survive across sessions.
In my view, that is the part enterprise technologists should examine first. Long-lived agents change the evaluation problem. Accuracy in a single answer is no longer enough. Teams will need to judge reliability over hours, handling of interrupted tasks, behavior when a connected application changes or denies access, and auditability of what was read, clicked, or sent while no one was watching. Computer use plus connected applications is powerful precisely because it touches systems of record.
Worth flagging for IT owners: availability scoped to Business Premium and Pro, with a higher-priced tier beside it, points to where OpenAI expects early load to concentrate. Business users with real calendars, real documents, and real permissions. That is also where the operational lessons will come fastest, around provisioning, offboarding a dot when an employee changes roles, and scoping what a proactive assistant is allowed to do unattended.
I have watched my own children treat persistent software as normal in a way my generation did not. They do not check on it. They assume it continues. If Dots works as described, office software may start to feel the same way. Work gets assigned, it proceeds, and the human returns for review. The technology arc, when it works, bends toward less drudgery. Launching a site, reconciling a schedule, drafting the board deck: none of that is creative in the best sense. Offloading it cleanly would leave more room for the decisions that actually require judgment.
