In Brief
- Automated team workflows — Workspace Agents handle reporting, coding and communication tasks continuously.
- Agents evolve through use — Because agents have memory and can be corrected in conversation, they improve over time and become a practical way to keep team knowledge current.
- Scheduling and deployment options — Agents can run on a set schedule or be deployed in Slack to pick up requests automatically, working even when users are offline.
OpenAI launched Workspace Agents in ChatGPT on April 22, marking its latest play for the enterprise following an extended push in 2025 and early 2026. With Workspace Agents, enterprise teams can create, share and manage AI agents for workflow automation. Underlying the agents is Codex — OpenAI's cloud-based coding agent — enabling them to handle reporting, coding and messaging tasks on behalf of teams.
The company called the release "an evolution of GPTs," which will remain available for the time being, with plans to phase them out in favor of agents.
The feature is in research preview for ChatGPT Business, Enterprise, Edu and Teachers plans. According to the company, Workspace Agents convert scattered team knowledge into reusable workflows with enterprise-grade controls. The agents are free until May 6, 2026, after which credit-based pricing applies.
Workspace Agents Features
The feature set reflects OpenAI's push to position ChatGPT as an operational work platform rather than a chat interface. Here's what's included at launch.
| Capability | Description |
|---|---|
| Workspace Agents | Cloud-based AI agents for collaborative workflow automation |
| Codex-powered execution | Agents write and run code, access files and memory |
| Natural language setup | Describe a workflow; ChatGPT builds the agent automatically |
| Permission controls | Agents request approval before sensitive actions like sending emails |
| Compliance API | Gives admins visibility into agent configuration and activity |
How Workspace Agents Work
Users create agents by describing a workflow in the ChatGPT sidebar, or by dropping in a file. ChatGPT then defines the steps, connects the right tools, adds skills and tests performance. Templates for finance, sales, marketing and other areas are available, each with built-in skills and suggested tools.
Agents currently integrate with ChatGPT and Slack, with the company stating more channels are on the way. They can be scheduled to run automatically or deployed in Slack to respond to incoming requests. Because agents have memory and can be corrected in conversation, they improve over time.
OpenAI shared a number of internal examples, including: a Third-Party Risk Manager that screens vendors for sanctions, financial and reputational risk, and an accounting agent that handles month-end close tasks including journal entries, balance sheet reconciliations and variance analysis.
One of our Sales Consultants built, evaluated and iterated a Sales Opportunity agent end to end without an engineering team. It researches accounts, summarizes Gong calls and posts deal briefs directly into the team’s Slack room. What used to take reps 5-6 hours a week now runs automatically in the background on every deal.
- Ankur Bhatt
AI Engineering, Rippling
The Compliance API gives admins visibility into every agent's configuration, updates and runs, with the ability to suspend agents if needed. The company states the workspace agents include built-in safeguards to protect against prompt injection attacks, but did not include any specifics of what that entails. An admin console view of every agent built across the organization, including usage patterns and connected data sources, is on the way as are automatic triggers, better performance dashboards and support for workspace agents in the Codex app.
OpenAI's Play for the Enterprise
OpenAI has made a sustained push into enterprise software throughout 2025 and into 2026, competing directly with erstwhile partner Microsoft, Google and Anthropic. The company added Connectors for Gmail, Google Drive, Outlook, SharePoint and GitHub in June 2025, then moved into enterprise search in October 2025 with GPT-5-powered "company knowledge" capabilities.
February 2026 brought Frontier, an enterprise platform focused on helping organizations manage AI coworkers with shared business context, execution environments, evaluation, and permissions VentureBeat — pitched against Microsoft's Agent 365, Salesforce Agentforce and Anthropic's Claude for Work.
Anthropic has pursued a similar enterprise strategy with its growing capabilities for Claude Cowork and Claude Managed Agents, although it has emphasized safety controls and transparency as differentiators in selling to large organizations. Both companies are competing for established enterprise software budgets held by Microsoft, Salesforce, ServiceNow and SAP, companies that have been automating business workflows for decades and are now integrating AI into their existing platforms.
Have a tip to share with our editorial team? Drop us a line: