88% of companies use AI. Only 33% have scaled it. The gap is not a technology problem. It is a discipline problem — and most vendors will not tell you that.

Buying an AI tool feels like progress. You have made a decision. You have a contract. The tool is installed. The dashboard is showing activity. You have done the thing.

Implementing AI is a different activity entirely. It requires defining what success looks like before you start. Training the AI on your specific content, policies, and brand voice — not a generic configuration. A staging review where a human reads what the AI will say to your customers. A process for updating the AI when your business changes. And someone accountable for whether it is actually working.

The Gap This Creates

PwC’s research says it plainly: technology delivers only 20% of an AI initiative’s value. The other 80% comes from redesigning work around the AI. Most businesses buy the technology and skip the redesign. Then they wonder why the ROI did not materialise.

The redesign is not complicated. It means deciding which queries the AI handles and which it always escalates. Updating your team on what changes when the AI goes live. Building the feedback loop where your team’s observations actually improve the AI. Reviewing results monthly and making changes based on data.

The Test

Simple test for whether you have implemented AI or merely purchased it: if the AI said something factually wrong to a customer today, would you know about it by tomorrow?

If the answer is no — you have purchased AI. If you have a weekly log review process, a named person responsible for it, and a clear path from “problem identified” to “problem corrected” — you have implemented it.