Note · · 2 min read
Most Businesses Don't Have an AI Problem
Before adding AI, make the decisions and handoffs in the actual work visible.
An internal app for a service business had to answer an ordinary question. What was happening on each job, and who could see it?
The work was specific. Supervisors used to-do lists, quality-control progress and project measurement sheets. The owner needed visibility across that work. After the build, he reported that the supervisors found the app easy to use and that he felt more in control of operations.
That is the result he described. I do not know how much time it saved. What I do know is that a useful system had to reflect the actual work closely enough for the people doing it to use it.
I think about this when the conversation turns to AI. A model can draft, summarise or suggest. It still needs a reliable account of the decisions the business makes. What counts as a completed quality check? Which measurement is the trusted one? Who reviews a score when something looks wrong?
Those are questions to answer with the people doing the work, not details to leave for a prompt to guess.
The first step can be quite small. Follow one job from start to finish. Notice where somebody has to explain the rule again, make an exception or carry information across a handoff. Make that logic visible, then decide where AI could help.
If three people described the same workflow in your business, would they describe the same business?