ChatGPT Stops Being a Chatbot: Inside OpenAI's New Spaces, Pages, and the Quiet War on Notion and Slack

At OpenAI’s DevDay 2026 in San Francisco on Tuesday, Sam Altman promised a product that would feel like a “whole new way of working with AI.” The headline grabber was Dots, the always-on personal agents. But the announcement that may matter most to the 1.2 billion people who use ChatGPT every week was quieter: ChatGPT Space, a shared team workspace — and the stack of collaboration features around it that turn ChatGPT from a conversation into something closer to an operating system for work.

Until now, ChatGPT has been fundamentally single-player. Your chats, your files, your context. Space changes that model. It is a persistent, shared environment where teammates, ChatGPT itself, and each person’s Dot work from the same pool of project knowledge — files colleagues upload, chats, generated artifacts — organized by ChatGPT according to rules the team defines. Simon Willison, live-blogging the keynote, described it as “shared artifacts, maybe Google Docs meets the artifacts pattern,” with a Notion-style slash menu driving the editing experience.

What Space actually does

In the on-stage demo, OpenAI’s Holly Li walked through building a fictional music app inside a Space. Her Dot, “Dottie,” had access to ChatGPT, Slack, Teams, and other connected tools, pulling context from shared materials to answer questions and hand off tasks to other agents. The vision is explicit: humans delegate, agents execute, and the shared Space is the substrate where all of it accumulates.

Availability is telling. Space ships first for Pro, Business, and Enterprise plans, on the desktop app and web, with mobile to follow. This is a workplace play from day one, aimed squarely at the budget line currently split between Slack, Notion, and Google Workspace — a collision course The Decoder’s Matthias Bastian laid out bluntly, noting it ultimately points at Microsoft Office and Windows too.

The supporting cast

Space did not arrive alone. Around it, OpenAI announced a family of collaboration primitives:

• Pages — a document type built for human-and-agent co-authoring. Users write, research, generate diagrams and images, and visualize data on a shared canvas, then invite teammates to edit.
• Collaborative Slides — multi-user, multi-agent deck editing, starting from a conversation or template, with comments and export to PowerPoint and Google Slides. Rolling out in the coming weeks.
• A Meetings plugin — listens to calls, generates notes, summaries, and action items, and files them into a Space. OpenAI says captured audio is deleted once notes are generated. Beta now on macOS for Pro and Business.
• ChatGPT in Slack and Microsoft Teams — an @ChatGPT mention in any channel, thread, or DM invokes the assistant, which can use admin- or user-approved connected tools. Business and Enterprise only.
• MCP Events and Team Tasks — the new MCP Events specification lets plugins trigger automations when something happens in a connected app: ChatGPT can watch a project board, read linked documents when tasks appear, and draft a plan unprompted. Team Tasks lets businesses delegate recurring work — weekly project updates, responding to inbound emails — to ChatGPT directly.

There is even a marketplace play: an OpenAI Marketplace launching with 32 partners — Adobe, Figma, Salesforce, ServiceNow, Harvey, CrowdStrike among them — where enterprises can redirect existing OpenAI spend toward approved partner software.

The strategic read

Two things make this more than a feature dump.

First, the timing is not innocent. OpenAI tried shared, multi-user chat before and shut it down in July to keep ChatGPT “simpler.” Reviving collaboration now — bigger and more agent-centric — is a direct response to where the market has moved in 2026: Meta’s Muse agent has sat at the top of the iOS free-app charts, Anthropic has pushed Claude deeper into enterprise workflows, and the workplace is where the actual money is. As Willison noted in his live blog, the Dots-and-Space UI “really does look like Meta’s Muse.” OpenAI is not inventing a category here; it is contesting one that formed around it.

Second, the architecture tells you where OpenAI thinks the next moat is. Every piece announced today — Space, Dots, the plugin system, “Sign in with ChatGPT” (which lets users spend their ChatGPT quota at 16 third-party services), the marketplace — converges on the same idea: ChatGPT becomes the identity, payment, and coordination layer for a broader agent ecosystem. The chatbot stops being the product and becomes the operating system.

The caveats

The live demo undercut the polish a little — “Dottie is having a slow morning,” an awkward silence, then a “still checking” status on stage. Anyone who has run agents in production knows the feeling. And the privacy questions are real: Spaces will inevitably hold sensitive company files, which is why OpenAI paired the launch with repeated references to its “Private Intelligence” system and its promise that business data stays protected. Whether that reassurance lands with procurement departments is a 2027 problem.

There is also the awkward fact that this DevDay landed hours after OpenAI cancelled its next frontier model, GPT-6.1 Astra, on safety grounds — a context that hung over the keynote even as the product news rolled out.

Still, the direction is unmistakable. The industry spent 2025 arguing about whether chatbots would replace apps. OpenAI’s answer in 2026 is that the chat was never the point — the workspace is.

Sources

• The Decoder — OpenAI’s reveals a new ChatGPT that looks less like a chatbot and more like an operating system
• Engadget — ChatGPT Space is a shared hub for work projects and personal AI agents
• Simon Willison — OpenAI DevDay 2026 live blog
• OpenAI — DevDay 2026 Recap
• VentureBeat — OpenAI launches Dots and ChatGPT Space
• CNBC — OpenAI DevDay 2026: Live updates and announcements