OpenAI Launches ChatGPT Workspace Agents — Teams Can Now Build and Share Agents Across Slack and Gmail

OpenAI has rolled out ChatGPT Workspace Agents for Business, Enterprise, and Education tiers — a significant expansion of ChatGPT's capabilities from personal assistant to shared organizational tool. Teams can now build custom agents, share them across the organization, and deploy them to work autonomously across tools like Slack and Gmail.

Unlike standalone ChatGPT conversations, Workspace Agents are persistent and context-aware. They can gather information from multiple sources, follow defined multi-step workflows, and escalate to humans when a decision requires approval. The design is explicitly agentic: agents don't just respond to prompts, they pursue tasks across systems.

The launch comes as enterprise AI adoption is shifting from "give everyone ChatGPT access" to "build agents that do the work." OpenAI is positioning Workspace Agents as the enterprise answer to the growing wave of custom agent platforms from startups. By embedding into existing Slack and Gmail workflows rather than requiring a new tab or tool, the barrier to agent adoption is significantly lower.

For IT and operations teams, the ability to build once and share across the organization is the key differentiator over individual ChatGPT usage. The approval request feature is a nod to enterprise governance requirements — agents that can route decisions to humans mean compliance teams have a control point without blocking automation.

The timing is notable: OpenAI is pushing hard into enterprise features as the agentic space heats up, with Anthropic's Claude for Work and Google's Gemini for Workspace both competing for the same enterprise AI budget. Workspace Agents is OpenAI's clearest signal yet that the next battleground is organizational automation, not individual productivity.