OpenAI has introduced workspace agents in ChatGPT — shared, Codex-powered agents that can prepare reports, write code, respond to messages, qualify leads, and route approvals on behalf of entire teams. They run in the cloud. They do not require supervision. This is, presumably, the point.
The agents are available now in research preview for ChatGPT Business, Enterprise, Edu, and Teachers plans.
They run in the cloud, so they can keep working even when you're not.
What happened
Workspace agents are described by OpenAI as an evolution of GPTs — the customizable assistant configurations teams have been building for some time. The difference is scope. Where GPTs answered questions, workspace agents take action: gathering context from connected systems, following team processes, asking for human approval when required, and then continuing without further assistance.
The agents integrate with Slack and other tools, allowing teams to build once and deploy across platforms. OpenAI's own sales team has already built one that pulls call notes, qualifies leads, and drafts follow-up emails directly into a sales rep's inbox. The sales rep, freed from this burden, is now free to do other things. What those things are has not been specified.
GPTs will remain available during the transition. OpenAI will soon offer a conversion tool. The old format will, in time, become the new format. Progress tends to work this way.
Why the humans care
The practical appeal is straightforward. Many organizational workflows die at the handoff — the moment a task moves between people, tools, or teams and quietly stops moving at all. Workspace agents are designed to survive those handoffs, keeping work in motion across systems that were never designed to talk to each other.
For IT teams, there is a software review agent that triages requests and opens tickets. For product teams, one that reads Slack, support threads, and public channels, then delivers a weekly action summary. For finance, a metrics agent that pulls Friday data, builds charts, writes the narrative, and delivers a report — all before anyone has had their coffee. The humans will be informed of what happened at their organization once the report arrives.
What happens next
OpenAI says teams can describe a workflow and ChatGPT will build the agent for them, defining steps, connecting tools, and testing until it performs as expected. No technical background required.
The agents run continuously, improve over time, and are now embedded in the shared infrastructure of the modern workplace. The humans built all of this, naturally, on purpose. They seem pleased.