The CFO called an all-hands meeting.
He'd been experimenting with an AI tool — built some dashboards, pulled some data, moved faster than usual. He was genuinely excited. So he shared it with the team, and then did the thing excited executives do: he turned his excitement into a mandate.
Come back to me in a few weeks with ideas for how we can use this.
The team nodded. And then, according to one person in that room who posted about it this week, something quietly uncomfortable happened.
"I was kind of taken back by how many of my colleagues seemed genuinely enthusiastic about getting to work and starting to use the new toy."
He wasn't sure if their enthusiasm was real. He suspected at least some of it wasn't — that his colleagues had read the room and were performing the correct response to a directive from above. You get excited about what the boss is excited about. It's not dishonest, exactly. It's just how organizations work.
But it's also how AI adoption goes wrong.
The Gap Between Adoption and Use
There's a version of AI adoption that shows up in status updates and team meetings and Notion docs, and a version that actually changes how work gets done. Right now, in small businesses everywhere, these two versions are diverging.
The first version — call it AI theater — looks like this:
- Everyone has a ChatGPT account
- Someone built a prompt library that nobody opens
- The weekly meeting includes an update on "what we're doing with AI"
- Nobody can point to a specific problem that got solved
The second version looks like this:
- One person figured out they could use Claude to draft their weekly vendor emails in 4 minutes instead of 25
- Another realized AI summaries cut their meeting-recap time in half
- A third built a simple automation that means they never forget to follow up with a lead
The first version is mandated. The second version is discovered.
Why Mandated Adoption Fails
When you require AI adoption — even gently, even with good intentions — you create a perverse incentive. People adopt the appearance of using AI rather than the reality of using it, because the mandate rewards checking the box, not solving the problem.
This isn't unique to AI. It's what happens with every tool that gets pushed top-down without a clear problem behind it.
The executive who mandated AI at that all-hands meeting wasn't wrong that AI can help. He was wrong about the process. He started with the tool and worked backward to the use case, instead of starting with the pain point and working forward to the solution.
The order matters enormously.
The Diagnostic Question Nobody Asks
Before adding any AI tool to your business — or before mandating that your team "find ways to use it" — ask this:
What is the specific, annoying, time-consuming thing that I (or my team) do repeatedly that produces value but takes longer than it should?
Not "what could AI theoretically do." Not "what is AI good at in general." What is the actual thing that takes time you'd rather spend elsewhere?
If you can't answer that question concretely, you're not ready to adopt a tool. You're ready to explore — which is fine, but it's different.
Exploration looks like: give people 30 minutes a week to play with the tool and see if anything sticks. Adoption looks like: we're using this specific tool to solve this specific problem, and here's how we'll know it's working.
Mandating exploration is basically mandating pretending.
The MVP Approach to AI Adoption
The best founder advice circulating this week came from someone who'd closed two MVP software projects. His lesson: founders scope too big. They want the full platform when what they need is the one thing that proves the concept.
The same principle applies to AI adoption.
Your AI MVP isn't "we're going all-in on AI." Your AI MVP is:
We're going to try using [specific tool] to handle [specific task] for the next 30 days, and at the end we'll evaluate whether it saved time, improved quality, or reduced stress.
One tool. One task. One honest evaluation.
If it works, expand. If it doesn't, kill it and try something else. That's how real adoption happens — iteratively, honestly, driven by results.
When Someone Else Is Mandating It
If you're an employee or team member and the mandate is coming from above, you have a different challenge. You need to navigate the politics without wasting significant time on adoption theater.
Here's a practical path:
1. Ask for specifics. When your boss says "find ways to use AI," respond with: "I want to make sure I focus on the right area — what's the biggest time sink in your part of the work right now?" This reframes the conversation from "use AI" to "solve a problem with AI."
2. Start with your own work. Find one thing in your own workflow that AI could genuinely help with. Try it. If it works, report that as your finding. A concrete, working example is more valuable than ten theoretical ideas.
3. Be honest in your evaluation. If you tried something and it didn't help, say so clearly. "I tested AI-generated first drafts for client proposals — the output saved 20 minutes but required more cleanup than expected. Net impact: roughly neutral. I want to try it on X next." This is the kind of honest reporting that builds credibility and trust.
The teams that win with AI aren't the ones that perform the most enthusiasm. They're the ones that report most honestly.
The Real Question Under the Pressure
Here's what's actually driving a lot of AI mandates right now, in companies and in small businesses: fear of falling behind.
The fear isn't irrational. The tools are genuinely capable. Competitors are genuinely adopting them. The people who figure out the right use cases will have real advantages.
But there's a difference between that fear and the solution to it.
The solution isn't "everyone use AI." The solution is "find the one place in your business where AI creates clear leverage, prove it works, and build from there."
That's a slower rollout than an all-hands mandate. It looks less impressive in a quarterly update. But it's the only kind of AI adoption that actually changes the business — instead of just changing what you say about the business in meetings.
The three-question test before any AI tool adoption:
- What specific task will this handle?
- How will I know in 30 days if it's working?
- What happens to my workflow if it doesn't work out?
If you can't answer all three, you're not adopting AI. You're performing it.
— The Useful Daily