ChatGPT Becomes an Office Suite: OpenAI's Space, Pages, and Collaborative Slides Explained

The suspicion that OpenAI is building a productivity suite is no longer a suspicion. At its fourth annual DevDay in San Francisco on September 29, 2026 — an event attended by 2,500 developers and packed with more than 20 launches — OpenAI rolled out ChatGPT Space, a shared team workspace whose main format, Pages, lets coworkers, ChatGPT itself, and the company’s new always-on Dots agents edit the same documents together in real time. Collaborative slides are close behind, arriving “in the next few weeks.” Together they turn a chatbot into something that looks and works a lot like Google Workspace or Notion, except every document comes with resident AI labor.

What Space, Pages, and slides actually do

ChatGPT Space replaces the old Library as “a new home for your team to collaborate with AI to get work done,” as OpenAI put it in its DevDay recap. It is available immediately to Pro, Business, and Enterprise users on the web and desktop apps, with mobile users able to find, read, and share pages for now (editing on mobile is promised soon).

Pages are the center of gravity. OpenAI describes them as “a format native to ChatGPT that AI can read and create, and that teammates can edit together in real time.” In the ZDNET demo, the interface resembles a Notion landing page — but with a crucial difference. During the on-stage demonstration, an OpenAI representative tagged her Dot agent directly in page comments, assigned it feedback from a coworker, and let it iterate on the project without anyone leaving the workspace. Pages can also update themselves from connected tools, which means a project page wired to your stack can stay current without a human copying data around.

Collaborative slides extend the same model to presentations. Multiple humans and multiple agents can edit a deck simultaneously — as in Google Slides — while ChatGPT handles on-page research, graphics, and image generation. Users can design conversationally or import their own templates, present directly from ChatGPT, and export to PowerPoint or Google Slides, allegedly without breaking formatting.

The deeper design point is where the work happens. “Today, people develop ideas in ChatGPT, move the work elsewhere to share it, then bring feedback back to ChatGPT,” an OpenAI representative told ZDNET. “Space brings that process together so people can create, share, and keep working with your team and ChatGPT in one place.” That loop — chat, export, share, re-import feedback — is exactly the friction a suite is supposed to eliminate.

Not launched in a vacuum

Space did not appear from nowhere, and OpenAI is not the first mover here. Anthropic recently shipped Claude Docs as part of its “Claude Cowork” consolidation, which folded Claude Design and Slides into the same agentic home base and let multiple teammates, human and agent, collaborate on shared projects. ZDNET’s Radhika Rajkumar notes that for both companies, “document and slide creation are now features inside a central agentic home base rather than disparate tools.”

OpenAI’s own trajectory tells a similar story. A year ago DevDay was about apps inside ChatGPT and an agent builder; since then the company launched ChatGPT Work in July as an answer to Claude Cowork, and now Space, Pages, and slides complete the picture. Sam Altman presented the pieces as a coherent new system on stage, though notably he did not frame it as a Workspace competitor — the company is hesitant to claim suite ambitions outright, even while shipping the components of one.

The rest of the DevDay drop reinforces the workplace push. Business and Enterprise users can now form teams inside ChatGPT and set up Team Tasks that run on a schedule or trigger on events like a new Slack message. Users can @-mention ChatGPT in Slack and Microsoft Teams — and teammates don’t need their own ChatGPT license to use it. A Meetings plugin takes notes on calls and in-person conversations without a bot joining (currently in beta on the Mac desktop app, with an explicit consent requirement and audio deletion after notes are generated).

The enterprise trust layer

The quiet announcement that may matter most for adoption is Private Intelligence, a data-access approval layer for enterprise customers. “Zero Data Retention with Private Safety Processing enables automated safety reviews without giving OpenAI personnel access to the underlying content,” the company explained. A Private Inference preview arriving this fall promises confidential computing with “strict, verifiable controls.”

The name is immediately reminiscent of Apple’s Private Cloud Compute, and it targets the single biggest blocker to enterprise AI adoption: the fear that sensitive data leaks to agents or to the lab behind the model. Glean’s CISO Sunil Agrawal offered a supporting quote tied to OpenAI’s no-training commitment and zero data retention. For companies being asked to move document collaboration itself into ChatGPT, this is the price of admission.

Who pays, and how much

Access to Space follows the money. It ships on Pro, Business, and Enterprise plans — which conveniently got a restructure at the same event. A new $500/month Pro 500 plan is the only tier with “Ultrafast,” generating up to 300 tokens per second in Codex (up to eight times standard speed). Meanwhile the existing $200 Pro plan keeps its price but, from October 30, sees its included Work and Codex usage halved from 20× to 10× the Plus allowance, and weekly GPT-6 Pro messages cut from 200 to 100. Existing subscribers keep current limits until October 29 and receive a one-time $2,500 usage credit grant expiring at year’s end. OpenAI has also committed not to bring back Pro 200’s five-hour limit.

The commercial context is hard to ignore: ChatGPT claims 1.2 billion weekly users, with over 35 million people building with Work and Codex weekly, while Anthropic has dominated enterprise revenue. Turning the consumer chatbot into a team workspace — with Slack and Teams integrations, a plugin marketplace (early partners include Canva, Figma, and Adobe), and Sign in with ChatGPT flowing plan usage into 16 partner tools including Notion, Devin, and Vercel — is OpenAI’s bid to convert reach into enterprise seat revenue.

The real contest: context ownership

Beneath the feature list, Space is a play for where work context lives. If a team’s documents, decks, meeting notes, and connected-tool state all reside inside ChatGPT — readable, editable, and continuously acted upon by resident agents — then the AI doesn’t just assist the workflow; it hosts it. That is precisely the territory Notion, Google, Microsoft, and Anthropic are contesting, and MCP Events, OpenAI’s newly proposed automation standard, suggests OpenAI intends to define the plumbing too.

The open questions are the usual ones for agentic workplaces: what happens when a Dot edits a page incorrectly at 3 a.m., who audits agent actions in regulated environments, and whether enterprises will trust a chat company with their document layer before they trust it with their code. But as a statement of direction, DevDay 2026 was unambiguous: OpenAI is no longer just answering questions. It is asking for your team’s shared workspace — and it shipped the product to prove it.