AIToday
Large Language ModelsAI Business & IndustryZenn AI/MLPublished: Sep 30, 2026, 22:00 JST

OpenAI's September 29, 2026 DevDay debuts always-on agent dots

OpenAI's September 29, 2026 DevDay debuts always-on agent dots

3 Key Points

  1. What happened

    At its September 29, 2026 DevDay, OpenAI announced more than 20 items, including dots, an agent running on GPT-6 Astra with its own cloud PC that works between conversations.

  2. Why it matters

    An agent that keeps working without being prompted could shift AI from answering single questions to carrying ongoing responsibility, though OpenAI still says errors remain and asks users to check important output.

  3. What to watch

    Dots is included in eligible plans at no extra cost, but starting or managing Work and Codex tasks still consumes normal ChatGPT usage limits, so it is not unlimited production.

WHO IT HITSEditorial and production teams who currently re-explain context across tools, and ChatGPT Pro, Business Premium, Enterprise, Edu and Healthcare administrators who must enable it.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

DevDay 2026 centered on the burden of "glue work": re-explaining background in a new chat, hunting for which drafts a document update affects, and redistributing edits by hand. The three items highlighted here line up in a chain — dots keeps working between conversations, Space holds the shared artifact, and Sol supports execution.

The dots design separates "always running" from "changing anything on its own." Unprompted research touches connected apps only through read-only tools, and sending, editing, or operating a browser or PC is off limits in that mode. Tasks it actually runs pass through auto-review against user instructions, Custom Rules and built-in safety requirements. OpenAI says the first dot is included in eligible plans at no extra cost, but starting or managing Work and Codex tasks still consumes normal usage limits, so it is not unlimited production.

What remains unmeasured is whether this cuts real working time or performs well in Japanese. How much these announcements change day-to-day work is likely to hinge on whether the shared Spaces and the promised automatic updates arrive as announced, and how much checking users still have to do.

FAQ
Who gets dots, and does it cost extra?
Pro is covered except in the EEA, Switzerland and the UK, and Business Premium is also eligible. The first dot is included in eligible plans at no extra cost, while Enterprise, Edu and Healthcare get it as an admin-enabled beta.
Can dots send messages or change things on its own?
Its unprompted research can only read connected apps with read-only tools, and it cannot send, change content, or operate the browser or PC in that mode. Tasks it does run go through auto-review against user instructions, Custom Rules and built-in safety requirements.
Does dots use up my ChatGPT limits?
Conversations with your dot are not counted against ChatGPT usage limits. However, Work and Codex tasks you start or manage from it do consume those limits as usual.

Also reported by Impress Watch, Latent Space, OpenAI Blog

AI news that matters for your work, in one minute a day

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleDazzle AI coach tells men: no contact won't win her back