OpenAI shipped workspace agents in ChatGPT on April 22, 2026 — shared, Codex-powered agents that run in the cloud and keep working after you close the tab. They’re available now in research preview for Business, Enterprise, Edu, and Teachers plans.
Workspace agents are a direct evolution of GPTs, but with meaningful differences. They run persistently in the cloud, connect to external tools, hold memory across sessions, and can be deployed into Slack so they pick up requests as they arrive. OpenAI’s own teams are already using them: the sales team built a lead outreach agent that pulls call notes, qualifies leads, and drafts follow-up emails directly in reps’ inboxes. The accounting team built one that handles month-end close — journal entries, balance sheet reconciliations, variance analysis — in minutes, with workpapers generated automatically.
Five pre-built templates cover the most common use cases out of the box: software request triage, product feedback routing, weekly metrics reporting, lead outreach, and third-party vendor risk screening. Each comes with built-in skills and suggested tool connections. Admins retain full control — they can restrict which tools and data each agent can access, require human approval before sensitive actions like sending emails or editing spreadsheets, and monitor every agent’s configuration and run history through the Compliance API.
IMPORTANT
GPTs are not being deprecated immediately. OpenAI confirmed GPTs stay available while teams test workspace agents, and a GPT-to-workspace-agent conversion tool is coming soon.
To get started, click Agents in the ChatGPT sidebar, describe a workflow, and ChatGPT walks you through building it. If your team already has GPTs doing useful work, hold off on rebuilding them manually — wait for the conversion tool.