OpenAI announced on Wednesday the launch of an entirely new category of cloud-based artificial intelligence tools called “Workspace Agents” within the ChatGPT platform. These innovative tools are specifically designed to automate complex workflows across various enterprise and corporate teams. Powered by the company’s Codex technology, this new feature represents what OpenAI describes as a radical evolution of GPT models. The feature is immediately available as a research preview to subscribers of business, enterprise, and educational plans, as well as educators.
This move is not merely a software update, but a redefinition of how humans interact with machines in modern work environments. Companies have long faced significant challenges in connecting disparate systems, often relying on intensive human intervention to transfer data or execute routine steps. With the arrival of these new agents, these operational bottlenecks are eliminated as artificial intelligence manages end-to-end workflows independently and intelligently. Furthermore, the company highlights its commitment to developing technologies that adapt to business needs by providing a secure environment that ensures enterprise data confidentiality—the primary concern that previously hindered the expanded use of AI in sensitive sectors. With this development, technology transitions from providing answers and consultations to actually executing tasks, unlocking new horizons for enterprise productivity and innovation.
The end of the custom model era and the beginning of the agent era
Workspace Agents are designed to replace the custom GPT models launched by the company last year. This upgrade marks a qualitative shift, featuring continuous cloud-based operation and deeper integration with daily work tools. In its official announcement, the company stated: “Teams can now build shared agents that handle complex tasks, long-running continuous workflows, and operate within the permissions and controls set by their organization.”
These intelligent digital entities possess broad capabilities, preparing detailed reports, writing code, responding to messages, and working across multiple platforms including ChatGPT and Slack. Because these tools operate in the cloud environment, they continue performing tasks even when users are offline, ensuring round-the-clock operational continuity. Thanks to this design, teams can build an intelligent agent once, share it across all departments of the organization, and subsequently improve and develop its performance over time.
OpenAI noted that these agents are capable of gathering data from multiple interconnected systems, strictly adhering to team protocols, requesting human approvals when needed, and maintaining consistency and harmony across various tools used. To ensure the highest levels of security and control, administrators in enterprise and educational plans can precisely govern the tools and actions available to different user groups. A compliance API has also been provided to closely and continuously monitor agent configurations and operations.
Free access and future pricing plans
In a move aimed at encouraging technology adoption, Workspace Agents will be available at no additional cost until May 6, 2026. Following this date, OpenAI intends to implement a credit-based pricing system. The company’s ambitions do not stop there; it plans to introduce automatic execution triggers, enhanced performance dashboards, and broader integrations with diverse business tools. The company also confirmed that it will allow users in the future to easily convert their current custom GPT models directly into Workspace Agents.
Massive expansion of Codex technology capabilities
This launch caps off a series of intensive activities by OpenAI, which has spent recent weeks expanding the capabilities of Codex technology with unprecedented strength. On April 16, the company released a major update branded “Codex for almost everything,” which added computer-use capability on macOS, an in-app browser, image generation capabilities, persistent memory, and more than 90 add-ons including tools like Jira, Microsoft 365, Notion, and HubSpot.
Earlier in April, OpenAI also introduced pay-as-you-go seat pricing exclusively for Codex within business and enterprise plans, lowering barriers for teams to test and trial the technology. The announcement of Workspace Agents is the latest clear signal that the company views enterprise adoption of autonomous agent-based AI as a foundational pillar and central element of its growth strategy—a direction clearly outlined by the company in an April 8 post titled “The next phase of enterprise AI.”
Frequently Asked Questions
What are OpenAI workspace agents?
They are new cloud-based artificial intelligence tools within the ChatGPT platform powered by Codex technology, designed to independently and continuously automate complex tasks and corporate workflows even when users are offline.
Will custom GPT models continue to work?
The new feature is designed to gradually replace older custom models, and the company will later enable easy conversion of older models into the new agents to keep pace with ongoing developments.
When will fees be applied to this new feature?
The feature will remain available to subscribers for free at no additional cost until May 6, 2026, after which a new credit consumption-based pricing system will be implemented.
