OpenAI now allows eligible US organizations to launch business agents on ChatGPT that can answer questions on their behalf, part of a wider push by the company to put autonomous AI inside the workplace.

The rollout targets organizations in the United States that meet OpenAI's eligibility criteria, and it extends capabilities the company has been building out across ChatGPT since mid-2025. For organizations, it moves agent technology out of the hands of individual users and into a managed, company-level setting — one that connects directly to the business software teams already use day to day.

What OpenAI Is Rolling Out

The core capability is simple: organizations that qualify can set up agents on ChatGPT that respond to questions on their behalf.

The capability builds on workspace agents, a feature OpenAI introduced on April 22, 2026, built to run on the company's Codex model. Workspace agents are aimed at teams on the Business, Enterprise, Edu, and Teachers plans, and their main purpose is to automate complicated workflows that involve several steps.

The agents can plug into third-party business software, including Slack, Salesforce, and Microsoft tools. They can also act on their own: agents can run on a schedule or kick off in response to triggers, including API calls, and they operate within the permissions an organization sets. That combination — existing business integrations plus scheduled or trigger-based operation — is what distinguishes these agents from chatbots that only respond when prompted, since work can begin without a person starting each run.

OpenAI points to use cases such as lead qualification and report generation.

Guardrails, Pricing, and the End of an Era for Custom GPTs

Workspace agents come with role-based access controls and monitoring tools, so administrators can decide who builds what and track what the agents do. The agents also include constraints on the actions they can take and approval checkpoints. In practice, that means a human can be required to sign off before an agent does something consequential.

On pricing, OpenAI is starting with a free trial window. Workspace agents launched as a research preview and will stay free until May 6, 2026. After that date, the feature moves to a credit-based pricing model, under which organizations will effectively pay for how much work their agents perform, rather than a flat fee. That structure ties an organization's costs directly to agent activity, making actual usage the key variable in what a deployment ends up costing.

Workspace agents are expected to gradually replace custom GPTs as the standard for organizations. Custom GPTs—the configurable chatbots many teams built for internal use—will keep working for individual users. OpenAI plans to phase them out for enterprise customers, moving businesses toward the more capable agent format. For teams that built internal custom GPTs, the shift sets up a migration from configurable chatbots toward agents that can carry out multi-step actions under organizational permissions.

How OpenAI Got Here

In July 2025, the company launched a general-purpose ChatGPT agent for individual users. That release established the idea that ChatGPT could take actions, not just generate text. The organizational rollout extends the same concept to teams, adding the permissions and oversight that companies expect. The May 6, 2026 move from free preview to credit-based pricing is the next fixed milestone in that rollout, and the pace of the enterprise custom GPT phase-out will be a factor in how organizations plan the switch.

Source: CryptoBriefing. This article was edited by Diego Almada Lopez. Original report published November 10, 2026: https://cryptobriefing.com/openai-chatgpt-business-agents-us-organizations/