OpenAI launches ChatGPT Spaces for human-agent collaboration

What happened

OpenAI has introduced ChatGPT Spaces, a shared workspace designed to let people collaborate more directly with ChatGPT and with Dots, the company's agent framework. The feature replaces the previous Library environment, repositioning ChatGPT's interface around collaborative, persistent spaces rather than a simple archive of past chats.

According to TechRadar, Spaces is built to support ongoing work between humans and AI agents in one place, rather than treating each conversation as a standalone, disposable session. The move signals OpenAI's continued push to make ChatGPT function less like a chat tool and more like a persistent workspace for ongoing projects involving both people and autonomous agents.

Why it matters

This is primarily a product and interface shift, and it is a meaningful one. By retiring Library in favour of Spaces, OpenAI is signalling that the future of ChatGPT is organised around shared, ongoing work rather than isolated prompts. For organisations experimenting with agentic AI, this matters because it changes the unit of interaction: instead of a single query-response exchange, teams and agents can now operate inside a common environment that persists over time.

For leaders building AI into service, operations or internal workflows, this points to a broader direction — agents like Dots are being designed to sit alongside human collaborators as ongoing participants, not just one-off tools invoked for a task. That has implications for how work gets structured, how handoffs between humans and agents are designed, and how trust and oversight get built into AI-assisted workflows.

The Renascence take

The headline feature here isn't the rebrand — it's the behavioural assumption behind it. OpenAI is betting that people work better with AI when the relationship feels continuous and collaborative rather than transactional.

Most coverage will frame this as a cosmetic swap of Library for Spaces, but the real story is a shift in the mental model OpenAI wants users to adopt: AI as an ongoing collaborator with shared context, not a search box you query and discard. That's a meaningful behavioural design choice — persistent, shared environments build a sense of continuity and trust that one-off chats never could. Any organisation layering agents into customer-facing or internal workflows should take the same lesson: the interface and interaction model shape how much people actually trust and rely on an agent, often more than the agent's raw capability does. Before deploying agentic AI, design the workspace and the handoff experience with as much care as the model itself.