An owner decides the business should be using AI. The next question arrives quickly: who actually does the work?

There are three good answers, and most businesses end up using more than one.

Learn it

Your people learn to direct AI on their own work. They brief it, check what comes back, and build small workflows for the tasks they repeat every week.

This suits the many small jobs spread across a team: the weekly report, the standard reply, the first draft of a proposal. Nobody outside your business knows those tasks as well as the people doing them, and the skill stays with your team when the tools change.

Build it with us

Your team builds, on your real work, with a guide alongside. The guide helps them choose what to automate first, shape the instructions and set up the checks. The team does the building.

This suits work that matters to the business and involves several people, where you want the capability inside the company at the end. The team learns by building something it will actually use.

Have it built

Someone builds the system for you, and your team owns it and runs it. This suits work that is too heavy, too central or too urgent to wait while a team learns: a system several departments depend on, or a process that runs thousands of times a month.

The process still has to be clear first. A build reproduces whatever version of the task it is given, so the first step is the same as in the other two ways: write the work down and agree on one version.

Three questions to decide

How often does it come back, and for how many people? Small tasks that recur for one person suit learning. Larger work shared across a team suits building it together.

Who should own the skill afterwards? People improve what they helped build. Work you want your team to keep improving is work they should help build.

Can it wait for a team to learn? Heavy, central or urgent work is the case for having it built. Plan who will look after it once it is running.

The three work together

A team that has learnt the basics briefs a build partner far better, because it knows what is realistic. A built system lasts longer when the people around it understand how it works. Each way makes the others easier.

A good first step: list the AI work you have in mind, and mark each piece learn, build together, or have built.

In your department

  • Operations & Delivery: The weekly status report is a task your coordinators can learn. The scheduling system several sites depend on may be one to have built, with a named owner once it runs.
  • Strategy & Planning: Sort the year’s AI ideas into the three ways before you budget, so training and build spend each go where they fit.
  • HR & Admin: When you plan AI training, separate the tasks staff should learn to do themselves from the systems the company should have built.