Stop building dashboards. Start building decisions.

Every time I sit in front of a new AI project, someone shows me a dashboard. Usually it is beautiful. Usually nobody uses it.

This is not a criticism of dashboards. It is a criticism of the thinking that produces them. We have spent thirty years building ever prettier windows into data, and we are not meaningfully better at deciding what to do.

The deliverable of a management system is a decision. Everything before that is cost.

The confusion

The confusion comes from treating information as the goal. Information is not the goal. Information is what you buy on the way to a decision. If the decision does not improve, the information was not useful.

A warehouse supervisor does not need a dashboard. She needs to know whether to call the second shift tonight. A CFO does not need a dashboard. He needs to know whether to push the quarterly forecast down by three percent. These are decisions, not displays.

AI lands inside organisations when it shortens the distance between a question and a decision. Not when it produces a more colourful chart.

What changes with AI

Two things change.

First, the cost of asking a question falls to almost zero. You no longer need an analyst and a week. You can ask the question and get an answer in the time it takes to type it. That is a real shift, but it is a shift at the bottom of the stack.

Second — and this is the one most people miss — the cost of sitting in front of structured ambiguity also falls. An AI will happily hold twelve factors in mind at once, weigh them, and give you a reasoned recommendation, knowing that it is a recommendation and not an answer. That is new. Humans cannot do this at scale. A good operations manager can, for one decision at a time. An AI can do it for a hundred, every morning, before you have your second cup of tea.

The move

The move, then, is not to put AI behind the dashboard. The move is to put AI in front of the decision.

Start with the decision the organisation actually makes. Write it down. Describe it the way you would describe it to a new person on their first week. What are the inputs. What are the considerations. What is the call. What would have to be true for the call to flip.

Then ask whether an AI, given that description and access to the organisation’s data, could make that call — or at least propose the call, with reasoning, so a human can accept or override.

That is a decision engine. That is a useful thing.

What this means for leaders

It means you stop buying software and start designing decisions. It means your job as a leader is to articulate what the organisation decides, not what it measures. It means the IT team and the operations team need to sit at the same table, and the IT team needs to stop being the last one in the room.

And it means, quietly, that a lot of consulting decks become obsolete — because the deck was a dashboard, and the dashboard was never the thing.

The thing was the decision.

That is where I now do my work.