The short answer.

AI changes a manager’s job by making the manager the owner of how the team’s work is done, in writing. Three things are new: you decide which pieces of the team’s work AI prepares, you keep the team’s method current as the work changes, and you look at prepared work and at how people check it, more than at who is busy. Setting the standard, developing your people and deciding the hard cases stay exactly where they were: with you.

Once a workflow prepares part of the team’s work in ChatGPT, Claude, Copilot or Gemini, the team’s method has to be written down precisely enough to run the same way every time. On a team, the person who owns that page is the manager.

Three things are new in a manager’s job.

A manager on a team that runs AI workflows takes on three responsibilities that did not exist before.

  1. You decide which pieces of the team’s work AI prepares.The manager chooses the work, one piece at a time, in an order: the repeating work first, the work the team likes least near the top. The choice goes on the company’s plan, under your team, so the rest of the company sees it.
  2. You keep the team’s method current as the work changes.A workflow is your team’s best way of doing one piece of work, written down. When a rule changes, a client asks for something new or a system’s export moves a column, the manager makes sure the written method changes too.
  3. You look at prepared work and at how people check it.The useful question used to be “who is busy?”. On a team with workflows, the useful questions are “is the prepared work right?” and “is it being checked against our standard?”. The manager reads a few prepared drafts each week, beside the checks that were made on them.

None of the three is a technical job. Who owns an AI workflow after the pilot? sets out the roles around one workflow; this piece is about the manager’s week.

What stays the same.

The core of a manager’s job does not move: you set the standard, you develop your people, and you decide the hard cases.

  • You set the standard.What good work looks like on your team is still your call. The written method is where that standard now lives, so the standard gets sharper.
  • You develop your people.People still grow by doing hard work with a good manager beside them. With routine preparation done by a workflow, more of your coaching goes to judgment: the exception, the difficult client, the case the method does not cover.
  • You decide the hard cases.A workflow prepares; a person decides. People and AI on one team: who does what sets out that split line by line.

The plan rates and ranks nobody. It describes work, never performance. Nothing in a workflow records or watches people: the exception log records cases, and the manager reads work, not activity.

What do I do differently on Monday?

On Monday you start the week with the team’s prepared and checked work, not with a status round of who is doing what.

  • Open the exception log first.The exception log is where the people who run a workflow note each case they stopped or corrected. A correction that appears twice is a gap in the written method, and closing that gap is the manager’s first job of the week.
  • Read prepared work beside its check.Pick three pieces of last week’s prepared work and read what the reviewer changed.
  • Bring the hard cases to the team.The cases a workflow flagged as unclear are the best teaching material a team has. Walk one through with the person who checked it.

What do I stop doing?

A manager on a team with workflows stops doing the work that the written method now does.

  • Stop correcting the same thing twice.When you correct a draft, correct the method. The next draft arrives right.
  • Stop being the only copy of the standard.If your team asks you how something is done, the answer belongs in the workflow, where everyone, including the next new hire, finds it.
  • Stop measuring effort by who looks busy.Prepared work and careful checking are the measures that matter. Busy is not one of them.

A manager’s weekly checklist to copy.

Copy these five lines into your calendar for the start of each week. They take the place of the status round, not an addition to it.

  1. Exception log.Which correction appeared more than once: ______. The change to the method it needs: ______.
  2. Prepared work.Three drafts read beside their checks: ______. Was each one checked against our standard: ______.
  3. What changed.A rule, a client, a system’s export or the AI platform: ______. The workflow it touches: ______.
  4. What AI prepares next.The next piece of the team’s work for the plan, or the current order confirmed: ______.
  5. One hard case.The case I walk through with a colleague this week: ______. With whom: ______.

Anything on lines one to three that needs more than an adjustment goes to the monthly AI review.

One manager’s week.

Illustrative example.

Take a 70-person insurance brokerage. The commercial account management team runs two workflows: one prepares each client’s renewal summary from the policy export and last year’s notes, and one drafts the certificate request reply for a person to approve. The team lead owns both.

On Monday the team lead opens the exception log. Two account managers corrected the renewal summary the same way: it listed a client’s vehicles from the old schedule after a mid-term change. The team lead adds one line to the written method, to read the latest endorsement first, and has the saved test cases run again before the team uses the new version.

Next the team lead reads three renewal summaries beside the account managers’ checks. One summary was approved without the coverage comparison being read. The team lead fixes the summary, not the person: the team lead moves the comparison to the top of the summary, where nobody skips it.

Line four of the checklist: the team agrees that the next piece of work for the plan is the claims status update clients ask for each week. Line five: the team lead walks a newer account manager through a renewal the workflow flagged as unclear, a client adding a new line of business mid-term, and decides that case together.

Every decision about a client stayed with people. What changed is where the team lead spent the week.

What is different in a company of under 200 people.

In a company of under 200 people, the manager is usually also the person who keeps the workflows current, because there is nobody else to hand that to. Most published advice assumes a central AI team and a training department; a smaller company has neither.

  • The manager often does the work too.A team lead who also runs the workflow sees its gaps first. Name a colleague to read your own prepared work, so you are checked too.
  • The written method is the training.With no training department, the workflow is how the team’s standard reaches a new hire. How new hires learn the AI workflows their role runs shows that first week.
  • One owner sees across teams.Each manager owns a team’s method; one person owns the company’s plan. Who should own AI in a company your size? names that person.

Where the team’s method lives.

A manager’s new responsibilities sit inside the company’s plan for AI: plan it, develop the workflows, deploy them. How organizations plan their use of AI sets out all three steps, with a one-page plan to copy.

FullyPowered Planner is where the plan lives. You set out your teams, their roles and the work between them, and choose what to build first. Each manager adds their own team and sees their part of the same plan. The Workflows Directory holds free workflow templates to start from, and we build any workflow your team needs in a 3-Day Workflow Sprint. The workflows run in the AI your company already pays for, on your own account, run by your own people.