You're paying for Maestro's AI. Here's why your planners aren't using it.

Ask a planning director which Maestro AI capabilities are live and you’ll get a list. Ask a planner which ones changed a decision last week and you’ll get a pause. The gap between those two answers is money already spent.

Why the gap exists

Nobody audited the licence against the estate. Features were included in a commercial bundle and never mapped to a planning process. The platform can sense demand; the demand planner still starts from last month’s forecast because that’s the workbook they were trained on.

The use cases were chosen by the vendor, not the planner. The demo was impressive and the pilot was generic. Planners are rightly suspicious of a recommendation engine that doesn’t know their product mix has a promotional calendar.

Trust was never built. This is the real barrier. A planner who has been blamed for a stock-out will not hand a decision to an algorithm they can’t interrogate. No amount of configuration fixes that; only a deliberate period of side-by-side running, where the AI’s suggestion and the planner’s judgement are both recorded and compared, does.

The sequence that works

1. Audit the licence against the estate: every capability you pay for, mapped to the process step where it could act, with an honest note on data readiness.
2. Prioritise two or three use cases by value and by the planner’s appetite — not the vendor’s roadmap. Scenario intelligence for a supply planner facing a constrained supplier often wins over demand sensing, because the pain is sharper.
3. Configure and run in shadow for a defined period. The AI proposes; the planner decides; both are logged.
4. Review the log with the planners, not with the steering committee. When the numbers show the suggestion would have been right more often than not, trust moves. When they don’t, you’ve learned something cheaper than a rollout.
5. Measure the decision, not the login: forecast accuracy, inventory turns, expedite spend, time to resolve an exception.

The honest caveat

Some licensed capabilities won’t earn their keep in your estate, and the audit should say so. “AI-powered” is not a planning outcome. A planner who trusts one automated workflow enough to leave it running is.

That’s the scope of our Maestro AI Enablement work — and, as with everything we do, it starts with an assessment you can act on without us.

What trust actually costs to build

Planners are not resisting AI because they misunderstand it. They are resisting it because they carry the consequence of being wrong, and the system does not.

A demand planner who accepts a sensed forecast and ends up with six weeks of cover on a slow mover will have that conversation with their director personally. The algorithm will not be in the room. Until the accountability and the automation sit in the same place, a rational planner discounts the suggestion — and that is not a training problem.

Three things move it, in our experience, and none of them is a workshop.

Shadow running with a written record. For a defined period — six to eight weeks is usually enough — the AI proposes and the planner decides, and both are logged against the same outcome. No pressure to comply, no dashboard of adoption rates. Just a record.

Reviewing the log with the planners, not about them. At the end of the period, sit down and count. Where was the suggestion better? Where was it worse, and why? The “why” matters more than the tally: planners will often accept a system that is wrong in ways they understand, and reject one that is right in ways they cannot explain.

Naming who owns the automated decision. If a workflow runs unattended, somebody other than the planner owns the outcome when it goes wrong. Write that down before go-live. This single sentence does more for adoption than any amount of change management.

Where the value usually is

Clients expect demand sensing to be the win. Often it isn’t — not because it doesn’t work, but because the demand planner’s pain is spread thinly across many SKUs, so an improvement is hard to feel.

The sharper pain is usually on the supply side. A supply planner facing a constrained supplier is making a high-stakes allocation decision under time pressure, repeatedly, with incomplete information. Scenario intelligence there is felt immediately: the planner runs three options in the time it used to take to build one spreadsheet, and takes a better decision in front of a commercial director who can see the working.

Start where the pain is sharp and the feedback is fast. Adoption spreads from a planner telling a colleague it saved them an afternoon — never from a licence utilisation report.

What the audit should tell you

A capability audit worth the name produces four columns: the capability, whether you are licensed for it, whether your data can support it today, and the process step where it would act. Most estates come back with a handful of capabilities that are licensed, data-ready and pointed at a real decision — and a longer list that fails on data readiness rather than on licensing.

That second list is the more useful one. It tells you the AI programme you actually have is a data programme, and it is better to know that in week two than in month nine.