OpenAI launches ChatGPT Spaces for human-AI collaboration

What happened

OpenAI has introduced ChatGPT Spaces, a persistent shared workspace that replaces the platform's previous Library feature. According to TechRadar, Spaces is designed to support ongoing collaboration between human users and AI agents built on OpenAI's Dots agent framework, moving ChatGPT away from a simple archive of past chats towards a standing environment where work can continue over time.

Rather than treating each conversation as a discrete, closed session, Spaces lets people and agents return to a shared project context repeatedly — building on prior work rather than restarting it. The change signals a shift in how OpenAI expects ChatGPT to be used: less as a query-and-answer tool, more as a continuing workspace that sits alongside human teams.

Why it matters

The move reflects a broader direction in applied AI: vendors are increasingly designing for sustained human-agent collaboration rather than one-off task completion. A persistent workspace implies memory, context retention and some notion of ongoing "state" — capabilities that make agents more useful for multi-step projects, but which also raise new questions about how organisations manage continuity, access and oversight when an AI system is a standing participant in work rather than a transactional tool.

For leaders evaluating AI adoption, this is a signal that agent frameworks are maturing from single-turn assistants into collaborators embedded in ongoing workflows. That shifts the design and governance conversation: it's no longer only about the quality of a single response, but about how well a shared workspace supports handoffs, accountability and trust over repeated interactions between people and agents.

The Renascence take

Most coverage will focus on the feature swap — Library out, Spaces in — but the more interesting story is behavioural. Persistent, shared workspaces change the psychological relationship between a person and an AI system: it stops feeling like a tool you query and starts feeling like a colleague you work alongside.

The real design challenge here isn't technical, it's experiential: once an agent has persistent memory of a shared project, users will start forming expectations of continuity, judgement and accountability that mirror how they treat human collaborators — and will be disproportionately frustrated when the agent falls short of those expectations. Organisations adopting agent-based workspaces like this should treat the "handoff" moments between human and AI contribution as a core service-design problem, not an afterthought, because trust in ongoing collaboration is earned or lost in those transitions, not in any single response.