I spent about four months reading articles about AI before I typed a single prompt that mattered.
Four months of newsletters, podcasts, LinkedIn posts about "the future of work," all of it accumulating in my head as opinion without evidence. I knew the arguments. I had no data. The thing I was reading about was sitting open in a browser tab the whole time, waiting.
So here's what I should have done in week one.
Open any AI chat tool right now — ChatGPT, Claude, Gemini, whatever you have access to — and try this. Take one real piece of work you're carrying today. A meeting you need to prep for. An email you've been avoiding. A decision you're turning over but haven't written down. Something live, not hypothetical, with enough detail that it actually matters to you.
Type it out as a situation, not a question. Don't ask "how do I write a good email?" Paste in the actual context: who you're writing to, what happened, what you're trying to achieve, what you're worried about getting wrong. Give it the real problem, with the texture it has in your head.
Then watch what comes back.
Two minutes, start to finish. You don't have to send anything. You don't have to use a word of the output. You're not handing over control; you're running an experiment.
Here's what I want you to notice — and this is the only thing I want you to notice.
Not whether the output is good. Whether it's useful as a starting point.
There's a difference. Useful as a starting point means: does it give you something to push against? A draft you'd rewrite is more useful than a blank page. A suggested framing you disagree with is more useful than no framing at all. The question isn't "did AI solve my problem?" The question is "did I just move?"
Most people, when they try this for the first time, feel one of two things. Either mild deflation — it's not as impressive as the hype suggested — or something subtler and more interesting: a small, involuntary jolt of forward momentum. The problem they'd been carrying for two days got externalized. It became an object they could look at and argue with, rather than a fog they were inside.
That jolt is what this is actually about.
There's a concept in behavioural science called present bias — our tendency to weight the immediate future so much more heavily than anything beyond it that we'll defer hard things indefinitely, even when we know they matter. "I'll start properly when I understand it better" is present bias wearing a productivity disguise. The reading is comfortable. The doing is uncertain. So we read.
The two-minute test is designed to be small enough to beat that. Not because it's trivial, but because two minutes genuinely cannot be the problem. If you have time to finish this article, you have time for this.
Now, the honest complication.
The first time I did something like this — pasted a real, live, messy work problem into an AI chat tool — the output was mediocre. Earnest, generic, structured like a business school slide. I nearly closed the tab.
What I didn't understand yet is that the first response is almost never the useful one. It's the draft of a conversation, not the end of it. You push back. You say "that's too formal" or "you've misread what I need" or "ignore the second point, focus on the first." The tool gets sharper as you give it more signal.
A prompt that produces a bad first response isn't a failed experiment. It's the experiment. It tells you where the gap is between what you meant and what you said — which, in my experience, turns out to be extremely useful information in general.
That's the real point I'm making here, underneath the two-minute framing.
AI doesn't reward passive exposure. Reading about it, watching demos, nodding along to podcasts — none of that builds the thing you need, which is an intuition for how to talk to it about your specific work. That intuition only comes from doing it badly a few times, noticing what happened, and adjusting.
This is what The AI Habit is built around. Fifteen minutes a day — not a course you complete and shelve, but a daily practice that builds the muscle through small, concrete tasks, each one real enough to create actual signal. First 14 days are free, which means you can find out whether it changes anything before you've committed to anything.
Try the two-minute test first. Then, if you felt even a small version of that jolt, you'll know why the daily practice is worth it.
https://harrysharman.com/projects/ai-habit/