In this session of Ask the CFO, I sat down with Ken Scales to talk about AI.

It’s everywhere right now. Every company feels like they need an AI strategy. Boards are asking about it. Leadership teams feel like they’re behind if they don’t have an answer.

But most of the conversations I’m having with business owners aren’t about strategy—they’re about pressure. It usually comes out like this: “What are we supposed to be doing with AI?”

So instead of trying to cover everything, we focused on what actually matters—where AI helps, where it doesn’t, and what most companies are getting wrong.

Start With the Problem, Not the Tool

One of the first things Ken said—and it’s something I’ve been saying for years in a different context—is that you don’t start with AI. You start with the business.

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

That’s the disconnect right now. People are starting with the tool and hoping it leads them to a solution. It doesn’t work that way.

If you don’t know what problem you’re solving, AI just helps you do the wrong things faster. And I see that happen more often than not.

Where AI Is Actually Showing Up

When I asked Ken where he’s really seeing AI help businesses today, he broke it into three areas. It’s a helpful way to ground the conversation:

1. Automation
   This is where most immediate value shows up. Document handling, data entry, moving information between systems—things people are still doing manually that don’t need to be.
2. Personalization
   This is what most people are experimenting with right now. Writing emails, creating content, supporting communication. Useful, but not transformational on its own.
3. Predictive & decision support
   This is where it gets more powerful. Using data to actually inform decisions. But this is also where most companies hit a wall.

And that wall is data.

The Data Problem Nobody Wants to Talk About

When I go into a business, one of the first things I look at is how their data is structured. More often than not, it’s scattered across systems, owned by different people, and not always consistent.

Ken said it clearly:
“Most of the time the data is in silos… and a lot of times we’ll try to help them consolidate that information.”

This is where a lot of AI conversations fall apart. Because if your data isn’t clean and connected, you’re not getting better insights—you’re just getting faster confusion.

Before AI does anything meaningful, you need:

• clarity on what data matters
• consistency in how it’s entered
• alignment across systems

That’s not exciting work, but it’s the work that makes everything else possible.

A Practical Way to Start

There was a moment in the conversation where we got into how to actually approach this without overcomplicating it. Ken’s answer was simple and practical:

“Start with a pilot program… what small things are you trying to achieve… then measure that success.”

That approach works because it keeps things grounded. Instead of trying to roll out AI across the company, you:

• pick one problem
• test a solution
• measure what changed
• decide if it’s worth expanding

He also emphasized something I’ve seen firsthand—this can’t live in one department. If you don’t involve people across the business, you end up building something that works in isolation but doesn’t hold up in reality.

It Still Comes Back to Process

This is the part that never changes, no matter what technology we’re talking about.

Process still wins.

I shared a simple example during the session. A client was dealing with reporting issues at month-end. They assumed it was a system problem. It wasn’t. Nobody owned a key step in the process. Someone had left, and the responsibility never got reassigned.

Once we assigned ownership, the issue went away.

Ken sees the same pattern from a different angle. When companies say they have a cash flow issue, he walks it backward:

“Why do you have a cash flow issue? We’re not getting invoices out on time… why not? We’re not getting the information we need.”

There’s almost always a chain. And if you don’t fix the chain, AI doesn’t help—it just moves the problem around faster.

This Is More About Behavior Than Technology

One of the most useful perspectives in the session came from Joe Kuntner, who joined us live.

He said:
“This is mostly about behavior changes than it is about the tech.”

That’s exactly right. The technology is moving too fast to keep up with perfectly, and that’s not the goal anyway.

The goal is to build a business that can adapt. Teams that are willing to experiment. Leaders who are willing to try something, see what works, and adjust.

That’s what’s going to matter over the next few years—not which tool you picked.

Where I Land on This

If you strip all of this down, here’s where I come out.

You don’t need an AI strategy.

You need better decision discipline.

You need to understand how your business actually runs, where the friction is, and what matters. Then you can bring in tools—AI included—to support that.

If you skip that step, you’ll spend time and money without much to show for it. If you get it right, AI becomes a multiplier.

That’s the difference.

If you haven’t watched the full session yet, it’s worth your time. There are a lot of real examples in there that go deeper than what I covered here, and it will give you a better sense of how to start thinking about this in your own business.