AI & Business
AI is not the theme
4 min read
Most companies adopt the tools and change nothing, then conclude the technology was oversold. The problem is almost never the model. It is the absence of a system around it.
There is a predictable arc inside companies adopting AI. Licences are bought, a few people become enthusiastic, output goes up somewhere unmeasured, and eighteen months later nobody can point to a decision that is being made differently. The conclusion drawn is usually that the technology was overstated.
It usually was not. What was missing is that the tool was adopted without a system, and a tool without a system is a subscription.
A tool nobody on the team can operate is not a capability. It is a subscription.
A system means four unglamorous things. A defined decision the technology is supposed to improve. A workflow that produces an input for that decision on a schedule, without anyone remembering to trigger it. A named person who owns the output. And a measurement that tells you whether the decision got better, which requires knowing what it looked like before.
Most AI initiatives skip all four and go straight to capability. That is why they feel exciting for a quarter and evaporate in the second one.
This is also why I do not sell AI as a theme. In experiences it operates underneath: sharpening who should be in the room, personalising a guest journey, turning an evening into a follow-up sequence that would otherwise depend on someone's memory of a conversation. In growth work it operates in the same position — content systems, campaign infrastructure, reporting an executive will actually read because it answers a question they actually have.
The test I apply before recommending anything: if this works, which specific decision gets made better, by whom, and how will we know? If the answer needs a paragraph, the initiative is not ready. If it needs a sentence, it is usually worth building.