AI fluency doesn't come from reading more about AI. It comes from using these tools on work you already understand and paying attention to what happens. Treating AI as something to "get into" or "figure out later" creates pressure and delay. A better approach is to treat AI like any other skill: practice it in small, repeatable ways and learn from the results.
Think of this as five simple experiments. One per week. No big initiative. No commitment beyond the work already on your desk. Each one is designed to show you something specific about how AI behaves, where it helps, and where your judgment still matters.
Week 1: Rewrite Something You Already Wrote
Start with something familiar. An email you sent last week. A project update. A client response. A routine internal note. Paste it into ChatGPT or Claude. Ask for a shorter version. Or a clearer one. Or a version with a different tone. The value here is in noticing what changes. What gets cut. What gets emphasized. What gets flattened. Sometimes the result is cleaner and more direct.
Sometimes it turns into generic language that strips out everything that sounds like you. Try this with two or three pieces of writing and compare the outputs. You'll learn how AI structures information, where your voice shows up, and what "clear" looks like to a system that only sees text.
Week 2: Ask About the Thing You've Been Nodding Along To
Pick a term you've been nodding along to for years. The acronym that appears in every deck. The technical concept that gets referenced without explanation. The regulation people cite as if it's obvious. Ask AI to explain it plainly. Ask follow-up questions. Ask for examples. Ask it to explain differently. There's no social cost here. No one wondering why you're asking now.
Usually, the concept itself isn't that complicated. It was just never unpacked clearly. You'll learn where your knowledge gaps actually are and how much professional language is shorthand rather than complexity.
Week 3: Turn a Mess Into a Structure
This works best with things that feel unresolved. Meeting notes from multiple calls. A long email thread that never quite landed. Your own half-formed thinking about a problem you keep circling. Paste everything into AI and ask it to pull out themes, decisions, and open questions. You're not asking it to solve anything. You're asking it to show you what's there.
The structure it produces won't be perfect. It may group things you wouldn't. It may miss subtext you care about. That's useful. Reacting to a structure is easier than staring at a pile of information. You'll learn how AI spots patterns and where it lacks the situational context that shapes how you'd organize the same material.
Week 4: Generate Bad Options on Purpose
Blank pages slow people down more than hard problems. If you need a subject line, an opening paragraph, or a starting angle, ask AI for ten options. Most will be mediocre. Some will be wrong. A few will be close enough to react to. Take the least bad option. Rewrite it. Adjust the tone. Move forward. You're not outsourcing judgment. You're creating momentum.
You'll learn how to get unstuck quickly, how to work from rough material, and how little "perfect" matters at the start.
Week 5: Compare AI to Your Best Work
This is the most important experiment. Choose something you know well. A document you've written dozens of times. A type of analysis you're confident in. Ask AI for a first draft. Then compare it to your own work. What's missing? What sounds confident but misses the mark? What assumptions does it make that you wouldn't? This is where your expertise becomes visible.
Your value isn't producing text or analysis. AI can do that. Your value is knowing what matters, what doesn't, and why. You'll learn where judgment shows up, what experience adds beyond pattern recognition, and where the human part of your work lives.
Why This Works as a Practice
These five experiments are intentionally small. Nothing leaves your desk. Nothing requires approval. Nothing depends on perfect output. Each one teaches you something different about how AI behaves and how it fits into your work. Bad results still count. In many cases, they teach you more than good ones. Most importantly, this approach builds familiarity without pressure.
You're not trying to transform how you work. You're building signal through repetition.
What Fluency Actually Looks Like
After five weeks, you won't be an expert. But you'll have evidence. You'll know which tasks AI actually supports in your role. You'll recognize when output looks confident but lacks context. You'll ask better questions because you've seen what vague ones produce. You'll be clearer about where your judgment matters most.
What to Do Next
Pick one experiment. Do it this week with real work. Not when you have time. Not when you feel inspired. It may feel awkward. Awkward usually means you're learning something real. Then do another the following week. AI fluency isn't built in a sprint. It's built through small, deliberate practice with work you already care about. That's when it stops being a tool you're aware of and becomes a tool you're using.
Do this on your own or do it as a team. If you want the team version, I run Discovery Labs designed exactly for this. Message me.