The Operations Bridge

Point eight in the AI Strategy Points series. Last time was the intern metaphor. This post is about how AI initiatives actually earn the right to scale — and it isn’t through a big launch.

The instinct to launch big is usually wrong

Scaling is about successful test cases. That sounds obvious written down, but I still watch organizations skip straight to an enterprise-wide rollout because the technology looks ready and the leadership team is excited. The technology being ready and the organization being ready are two different questions, and the second one usually gets less attention than it deserves.

This is the same lesson John Kotter’s eight steps for leading change have been teaching for decades, applied to a new kind of initiative: you need short-term, value-added wins before you can credibly ask an organization to change at scale. AI is not exempt from that logic just because the technology feels novel.

What a good test case actually does

A well-chosen test case does more than prove the technology works. It builds a story people can repeat to each other — a specific team, a specific problem, a specific before-and-after that’s concrete enough to be believable. “We cut the time on this report from two hours to fifteen minutes” travels through an organization in a way that “we’re piloting an AI initiative” never will.

It also surfaces the real obstacles — the data quality issues, the workflow friction, the places where the tool’s assumptions don’t match how your team actually operates — while the stakes are still small enough that failure is a lesson rather than a crisis. Scaling before those obstacles are understood just multiplies them.

Choosing the right first win

Not every problem makes a good test case. I look for a few things: a real pain point that people already complain about, a team willing to tolerate some rough edges while the tool matures, a result that can be measured in a way everyone agrees on, and a timeline short enough that the win is still fresh when you go looking for the next team to bring on board.

Chasing the most impressive possible use case instead of the most winnable one is a common mistake. An ambitious project that stalls for six months teaches the organization that AI initiatives are slow and uncertain. A modest project that clearly works in six weeks teaches the opposite lesson, and that lesson is what actually enables scale later.

A question for leaders

Do you have a specific, repeatable success story right now that a skeptical colleague would find convincing? If not, that’s the gap to close before the next planning conversation about scaling.

Next: the difference between “what, so what, and now what” applied specifically to AI — and why that’s where the real definition of value comes from.