OpenAI is moving its enterprise and education clients toward a new model of automation with cloud-hosted 'workspace agents.' These agents, available within ChatGPT for Business, Enterprise, and education plans, are designed to execute multi-step tasks independently. Examples include a system that scours the web for product feedback and compiles a Slack report, or a sales assistant that drafts follow-up emails in Gmail.

The technical architecture represents a clear evolution from the company's earlier GPTs, which were largely conversational interfaces. Workspace agents are built to operate with greater autonomy, integrating with business tools, gathering contextual data, and requesting human approval at defined stages. They are shareable within an organization, allowing teams to deploy, use in ChatGPT or Slack, and iteratively refine a single agent.

This push into autonomous agents aligns with broader industry momentum, exemplified by tools like Anthropic's Claude Coworker and the viral OpenClaw project—whose founder now works at OpenAI. The launch signals a likely sunset for the GPTs platform introduced in 2023. OpenAI states GPTs will remain available during a transition period, with a migration path to convert them into the new agent framework. The engineering trade-off is clear: sacrificing the simplicity of a single-purpose chatbot for the complexity—and potential—of persistent, process-aware automation systems that operate across a user's cloud services.

Source: The Verge