This playbook comes from a ChatGPT Business rollout NetSys completed for a 51-person creative firm in New York City, across executive leadership, sales, marketing, finance and operations. It took 62 days and 79 project hours, ended with 9 live workflows and 4 shared work areas, and left the firm with an in-house AI champion and a quarterly training plan. The full project is in the NYC creative firm case study. The client's name is withheld.
A rollout plan here means the order in which teams join, how each group is trained, which workflows are built first and who keeps the work going afterward. The workspace settings that come before it are in our ChatGPT Business setup checklist.
Why roll ChatGPT out one department at a time?
Each department has its own documents, tasks and approval rules, and training lands better when it uses that material. In the NYC rollout, NetSys moved through the firm one department at a time and built each session around the work that team already did. A project manager coordinated the rollout and a Lead AI Implementation Officer led the implementation.
Going department by department also limits the blast radius of a wrong setting. If an app permission or a shared project is set up badly, one team finds it, it gets fixed, and the next team starts with the corrected version.
How do you group roles for ChatGPT training?
Group people by the work they do, not by the org chart. In the NYC rollout, executives trained one-on-one and administrative staff trained in groups of two to four. Grouping staff with similar responsibilities saved rollout time, because one session covered the tasks they had in common.
| Group | Format that fits | What the session should cover |
|---|---|---|
| Executives | One-on-one | Their own recurring work, such as briefings, correspondence and decisions they prepare for, and what they want their teams to use ChatGPT for. |
| Staff with the same responsibilities | Small groups of two to four | The tasks they share, practiced on the team's own approved documents and templates. |
| Pilot workflow owners | Working sessions with the builder | How the workflow runs, what to check in its output and how to report a problem. |
| New hires after launch | Onboarding with their department's material | The usage policy, the shared work areas for their team and the workflows they will use. |
The format column reflects the NYC rollout; the coverage column is how we plan sessions, and it changes with each team's work. Before scheduling, list every person with their department, role and the two or three tasks that take most of their week. People whose task lists match go in the same session.
How do you choose pilot workflows?
A pilot workflow is one recurring task that is built, tested and measured before it is offered to everyone. In the NYC rollout, NetSys worked with designated executives and staff to build and test priority workflows as agents or automations, and 9 went live. Shared work areas gave teams access to approved reference material and templates.
When choosing the first workflows, look for tasks that:
- Recur often enough that improvements show up within weeks.
- Have a named owner who will test the output and say whether it is right.
- Use material the team is allowed to put into ChatGPT.
- Have a reviewer for anything that leaves the company.
- Can fall back to the old way of working if the workflow fails.
Our ChatGPT workflow examples for business teams show candidate tasks by department. Start with fewer workflows than you think you need; a workflow nobody owns stops being used.
What does an internal AI champion do?
An internal AI champion is a staff member who becomes the first point of contact for ChatGPT questions inside the company. NetSys helped the NYC firm establish one in-house. In practice the role covers:
- Answering day-to-day questions so small problems do not wait for the next training session.
- Collecting requests for new workflows and passing the promising ones to whoever builds them.
- Reminding colleagues of the usage policy and where approved material lives.
- Spotting which teams are not using the tool and why, before the next refresher.
Pick someone who already does the work, is respected by colleagues and has time set aside for the role. The champion supports the rollout; administration of the workspace stays with the workspace owner.
How often should staff get refresher training?
ChatGPT changes often enough that one launch session is not enough. The NYC firm's support plan includes two separate one-hour training sessions each quarter to keep staff current. Use those sessions for features that changed, workflows added since the last session and questions the champion has collected. Our AI training for employees delivers role-based onboarding and these quarterly refreshers.
How should seats be planned across departments?
ChatGPT Business offers Standard and Premium seats, and a workspace can mix them, per OpenAI's billing and seats article (checked on November 13, 2026). In the NYC rollout, Standard seats made up 89% of the license mix and Premium seats the rest. Give Premium seats to the named people whose work needs the higher usage, and move seats as usage shows. Seat prices and setup budgets are in our ChatGPT Business pricing guide.
What does a department rollout plan look like on paper?
- Inventory. List each department, its people, their roles and their main tasks.
- Order. Decide which department goes first and which follow.
- Groups. Executives one-on-one; staff with similar responsibilities in small groups.
- Pilot workflows. Pick the first workflows with their owners, and build and test them before the department's sessions.
- Champion. Name the internal AI champion and agree how much time the role gets.
- Refreshers. Put the quarterly sessions on the calendar before launch.
- Review. After each department, record what to change before the next one starts.
Frequently asked questions
Should we roll out ChatGPT to everyone at once?
Usually not. Rolling out one department at a time lets each session use that team's own work and lets you fix settings before the next team joins. The NYC firm in our case study was onboarded this way.
How long does a ChatGPT rollout take for a company of about 50 people?
It depends on the number of departments, workflows and training groups. The 51-person NYC rollout in our case study took 62 days and 79 project hours, including 9 live workflows. Treat that as one example, not a quote.
Should executives be trained separately?
In our experience, yes. Executives in the NYC rollout trained one-on-one, which lets each session use their own correspondence, briefings and decisions instead of generic examples.
How big should training groups be?
Small. Administrative staff in the NYC rollout trained in groups of two to four, grouped by similar responsibilities so one session covered their common tasks.
What is an AI champion?
An AI champion is an employee who acts as the first point of contact for ChatGPT questions, collects workflow ideas and keeps colleagues aware of the usage policy. The workspace owner still handles administration.
How often should ChatGPT refresher training happen?
Plan for it every quarter. The NYC firm's support plan includes two separate one-hour sessions each quarter, which keeps staff current as features change.
Which workflows should we build first?
Recurring tasks with a named owner, approved source material, a reviewer and a fallback. Build and test them with the people who do the work before offering them to the department.
Sources
- NetSys case study: ChatGPT Business rollout for a 51-person NYC creative firm, linked above (project figures, client name withheld).
- OpenAI, Managing billing and seats in ChatGPT Business, checked on November 13, 2026.
How NetSys plans a department rollout
NetSys plans the department order and training groups with you, builds and tests the first workflows with the people who own them, helps you name an internal AI champion and schedules quarterly refreshers. The workspace itself is set up under our ChatGPT Business setup service, and the same department-by-department method carries into an Enterprise rollout.
Share your headcount, the departments involved and the workflows you want to tackle first, and discuss your ChatGPT Business rollout.
Discuss chatgpt business setup for your business.
Tell us about your current systems, the result you need and your timeline. We will discuss the work, responsibilities and pricing before you decide on an engagement.