Picture Monday morning. Every hiring manager with a new joiner opens the same capability, supplies the role and the start date, and reviews a first-90-days plan written to the standard of your best manager. One person owns it, knows who uses it, hears what it misses and decides what changes next. That is a people team doing the impossible with the people it has: 10X what it can do, on every new start.
Getting there is an operating responsibility, like closing the month or running payroll. It is the last step of how we plan: start from the outcome, map the work that crosses teams, land each capability on the roles that do it, prove it on real work, then roll it out. A workflow that passed its tests is ready for daily use. The rollout puts it there.
Give the rollout an owner.
The owner is the person accountable for the result the workflow produces: the people lead for onboarding plans, the controller for receivables review, the firm administrator for intake. They decide who uses it, approve every change and answer for its quality. IT owns access: the approved AI workspace, the permissions and the records the workflow reads. Name both before the introduction, plus who covers when the owner is away.
The rollout starts from a workflow already proved on real work. What we test before we call an AI workflow ready describes that proof, and pilot acceptance-test preparation builds the test your pilot runs first.
The owner’s rollout checklist.
Copy it, fill in the brackets and keep it with the workflow. Each step says what to do, who does it and what done looks like.
Rollout checklist: [capability name]
- Introduction. Show the team the job the workflow does, the inputs it needs, one finished output and the decisions that stay with people. Run one real case live. Who: the owner, at a meeting the team already holds; IT confirms access beforehand. Done when: every user has seen it run and knows where the current version and its job aid live.
- First use. Each user runs the workflow on one of their own real cases in the first week, with the job aid open and a named experienced user beside them or on a call. Who: each user, with that experienced user. Done when: every user has completed one case and had their questions answered.
- Review. Set who reviews each output before it moves on, against which checklist, and how long full review lasts before spot review. Who: the owner sets it; the named reviewer applies it. Done when: the reviewer has the checklist in writing and every output so far has been reviewed.
- Exceptions. Write down the cases where a user stops and asks, and whom they ask. Keep one exception log: the case, what the workflow did and what the user did instead. Who: users record; the owner reads the log every [week]. Done when: the stop list is in the job aid and the log has entries or a recorded “none this week”.
- Support. Name one place for questions, the person who answers and how fast. Who: the owner or a named experienced user; IT for access problems. Done when: every user knows where to ask, and every question so far has an answer.
- First follow-up. Meet [two to four weeks] after the start. Read the exception log, the reviewer’s notes and what users want changed. Decide what to change, what to retest and whether to widen use. Who: the owner, the reviewer, two or three users, and IT if access changed. Done when: the decisions are written down, every change has been retested and the next review date is set.
One team’s completed checklist.
An invented team filled in the same checklist to roll out the employee onboarding plan capability to its hiring managers.
Illustrative example with invented data. An invented 120-person engineering firm. The plan the owner wrote, before the start date.
Rollout checklist: employee onboarding plan
- Owner and access
- The people lead owns the workflow. The IT manager gives hiring managers and the people coordinator access in the firm’s approved AI workspace.
- Introduction
- Week 1, Monday, at the managers’ weekly meeting. The people lead prepares a plan live for a real new joiner in project delivery, then shows what stays with the manager: goals, introductions and the first assignment.
- First use
- Every hiring manager with a joiner starting this quarter prepares that joiner’s plan in week 1 or 2, with the people coordinator on a call.
- Review
- The people coordinator checks every plan against the onboarding standard before the manager shares it. Full review continues until the first follow-up.
- Exceptions
- Stop and ask the people lead about a role with no agreed responsibilities, a joiner returning from leave, or any question about pay or contract terms. Each one goes in the onboarding tracker.
- Support
- Questions go to the people coordinator, answered within one working day. Access problems go to the IT manager.
- First follow-up
- Week 4, Friday. The people lead, the coordinator and two hiring managers read the tracker and decide what changes before the next intake of graduates.
The workflow prepares each plan. The hiring manager and the new colleague shape and agree it, and the people lead decides every change to the workflow.
Keep it good after the first month.
After the first follow-up, the owner turns repeated misses in the exception log into new test cases and retests before any change reaches the team. New hires learn the workflow as part of their own onboarding. When the work changes, the workflow changes with it, through the owner.
The people who use it daily know best what to improve. Build your AI plan with the people who do the work shows how to ask them and how to report back on their suggestions. The first 90 days of an AI adoption plan places this rollout inside the wider plan, from naming an owner to deciding whether to expand.
The rollout runs on the platform your team already has: ChatGPT, Claude, Copilot or Gemini. You set the pace, and your organization keeps control of how the work gets done.
Bring the workflow you want your team using next month and the role that will own it.