AI Fluency > AI Tools

Every enterprise I talk to wants to know which AI tool to buy. That's the wrong question. The right question is whether your people know how to think with it once it arrives.

I built an AI function inside a global investment bank from zero. No dedicated budget. No mandate from the top. What I had was a thesis: the tool doesn't matter nearly as much as the fluency of the people using it. Two years and 98% voluntary adoption later, I'm more convinced of that thesis than when I started.

Here's what I mean by fluency. A tool-trained person opens a chatbot, types a question, gets an answer, and calls it a day. A fluent person looks at their entire workflow and asks: what decisions am I making, what information am I synthesizing, and where can AI reshape the process end to end? That's a fundamentally different posture. One is using a search bar. The other is redesigning how work gets done.

I've watched this play out across 8 practice groups and 11 offices. The teams that jumped furthest weren't the ones with the most technical backgrounds. They were the teams where people understood what AI could do conceptually, and then had enough structured practice to develop instincts. They stopped asking "can AI help with this?" and started defaulting to "how should I use AI here?" That shift in default posture is what fluency looks like, and it's worth more than any feature comparison.

This is also why vendor selection is less strategic than most companies think. The foundation models are converging. The major providers are all getting better at the same things at roughly the same pace. If you pick a strong model today, you'll have a strong model tomorrow. But if your workforce doesn't know how to prompt well, how to verify outputs, how to chain AI into multi-step workflows — it doesn't matter which model you chose. You've bought a race car for people who don't have licenses.

Fluency is also what compounds. A tool advantage is temporary. Someone else can license the same platform next quarter. But a workforce that has internalized AI-native thinking — that has built the muscle memory for decomposing problems, structuring prompts, validating outputs, and iterating — that workforce gets faster every month. Each person discovers new applications. They teach each other. The 40% of our user base that became power users didn't get there because we upgraded the software. They got there because fluency begets fluency: once you understand the thinking pattern, you see applications everywhere.

When I train teams, I spend very little time on interface walkthroughs. Instead, I focus on three things: how to break a complex task into AI-addressable components, how to evaluate whether an output is actually good, and how to build on a result rather than accepting it at face value. Those three skills transfer across any model, any provider, any product generation. They're durable in a way that tool-specific training isn't.

I built a skills library of 40+ structured workflows for exactly this reason. Each skill isn't about clicking the right button — it's a reusable thinking pattern applied to a specific type of work. When someone learns the skill for, say, synthesizing research across multiple sources, they're not learning a feature. They're learning a method. And that method works whether the underlying model is from this year or next year.

The uncomfortable truth for procurement-driven AI strategies is that the moat was never in the license. It's in the people. Two organizations with identical tool stacks will produce wildly different results based on whether their people are fluent or merely equipped. I've seen it firsthand: 83% sustained active usage in our deployment, not because we have a better tool than our competitors, but because we invested in making people genuinely good at using it.

If you're building an AI strategy and your first move is a vendor bake-off, slow down. Figure out your fluency plan first. Decide how you'll teach people to think differently, not just click differently. The tools will keep getting better on their own. Your people won't — unless you invest in them deliberately.