There’s a question I keep hearing right now, and it usually comes up pretty quickly in a conversation:

“How should we be using AI in our business?”

It sounds like the right question. It feels practical. It feels like something you shouldhave an answer to right now.

But it’s actually the wrong place to start.

In last month’s Ask the CFO session on AI with Ken Scales, we kept coming back to something much simpler. Ken Scales said it directly:

“What are we trying to achieve first? It’s not, let’s use a tool or use AI to fix everything.”

That’s the part most businesses are skipping. They’re jumping straight into tools—trying to figure out what to automate, what platform to use, what their “AI strategy” should look like—without first stepping back and looking at how the business is actually running. When you skip that step, you don’t get better outcomes. You just get more activity. More tools, more dashboards, more things to manage—and not much improvement underneath it.

If you want AI to actually help your business, you have to start one level deeper. You have to start with where the friction is.

Where is the business not running as well as it should? Most of the time, you don’t need a full analysis to find it. If you sit with it for even 20–30 minutes and walk through your operations, a few things will usually stand out.

Here’s a simple way to do that:

• Where are we consistently slower than we should be?
• Where are people doing repetitive, manual work?
• Where do we have to pull information together just to understand what’s going on?
• Where are decisions being made with incomplete or delayed information?

You don’t need perfect answers. You’re just looking for patterns.One or two areas will stand out pretty quickly. That’s your starting point—not a tool, not a platform, not a strategy document.

From there, you can ask a much better question: Is there a way to improve this—and does AI play a role in that?

That’s a very different conversation than starting with the tool. Most companies reverse that order. They pick the tool first and then go looking for a use case to justify it. That’s why so many of these efforts stall out or never quite deliver what people expected.

Next week, I’ll build on this, because even when you identify the right problem, there’s another layer that tends to get in the way—and it has nothing to do with technology.