Set up your workspace

The workspace: your enterprise AI environment

A ChatGPT Enterprise workspace is a dedicated environment for your organization. It centralizes member management, billing, and security, keeping conversations, GPTs, and files separate from personal use and governed by organizational rules. Employees who also have personal accounts keep the two environments strictly separate unless they merge their personal account into the enterprise one—a permanent action that cannot be undone.

Initial setup takes a few hours. In contrast, fixing a poorly planned access structure can take months, as you’ll need to audit and correct inconsistent permissions member by member. This is one of the few decisions in this course where it pays to move slowly.

Organizing access: groups, roles, and projects

In an Enterprise workspace, access control relies on groups and roles. Groups cluster people by team, function, or responsibility. Groups grant no permissions on their own; you assign roles to them and use them to share GPTs. Custom roles then govern access to specific features—models, projects, memory, ChatGPT Work—for each population.

Day-to-day teamwork happens in shared projects, which bring together conversations, files, and instructions around a specific work object. Marketing might work in a “Content Creation” project, HR in “Job Description Drafting,” and Finance in “Budget Preparation.” Cross-functional initiatives, such as a product launch, follow the same pattern: a single shared project for all stakeholders, rather than creating a new workspace where you’d have to rebuild everything from scratch.

Step-by-step configuration

OpenAI recommends preparing the workspace in this order before inviting users at scale.

1. Confirm roles, seat types, and the launch scope.
2. Verify domains and configure identity, reviewing automatic account creation settings.
3. Set up SSO and SCIM.
4. Create groups and assign custom roles.
5. Configure workspace settings: name and branding, workspace instructions, the AI policy presented to members, ChatGPT Work access, and default models.
6. Enable apps, GPTs, plugins, and approved domains.
7. Open ChatGPT Work, workspace agents, and Codex to the teams that need them.
8. Configure security and compliance controls, including the data retention policy.
9. Monitor adoption and implement spending controls suited to your billing model.

Settings that matter

The Permissions & roles section in workspace settings holds the key controls: available models, GPT creation and sharing, projects, shared memory and enhanced memory, link sharing, and local or cloud access to ChatGPT Work. Each setting applies either to the workspace’s default role or to a custom role, letting you, for example, reserve a feature for a pilot team.

The AI policy deserves special attention. Drafted in general settings, it displays to members every thirty days and whenever updated. This is the natural place to reiterate usage rules, list prohibited data types, and provide a contact for questions. Without clear guardrails, employees will spend their first few weeks replicating their personal account habits in the enterprise environment.

Apps and plugins are disabled by default in new Enterprise workspaces, except for a curated selection. Review this list carefully and enable each connector only after verifying what data it can read and write. Owners can also allow or block account switching, which lets users stay logged in to both a personal account and their work account simultaneously.

The most common mistake is inviting everyone before establishing groups and roles. As OpenAI emphasizes, administrators then end up cleaning up inconsistent permissions, and some members receive more or less access than intended.

Your turn

Map out your access model. List the groups you will create, the custom roles they will need, and for each role, which features to enable or disable: models, projects, memory, ChatGPT Work, and apps. Then draft a ten-line AI policy that will display to members, and have it approved before inviting anyone.

Official sources: ChatGPT Enterprise admin quickstart · Managing workspace settings in ChatGPT Enterprise · What is a ChatGPT Enterprise workspace? How can I switch workspaces? · Admin controls, security, and compliance for plugins and apps · Projects in ChatGPT (OpenAI Help Center)

Key takeaways

• An Enterprise workspace separates organizational content from personal use and governs it under company rules.
• Access is organized through groups and custom roles; teamwork happens in shared projects.
• Follow the recommended order: roles and seats, identity, SSO and SCIM, groups and roles, settings, then feature enablement.
• The AI policy displays to members every thirty days: use it to reinforce your usage rules.
• Do not invite users at scale until you have established groups and roles.