I nearly called it a course. Then I remembered every online course I've ever bought and not finished, and I changed my mind.
The AI Habit is not a course. The distinction isn't marketing. It's the entire point.
Here is the problem I was trying to solve. Most people who want to get better at AI do something like this: they read a tweet thread listing "10 prompts that will change your life", try two of them, get middling results, and quietly file AI under "not quite for me yet." The information was there. The intention was real. Nothing stuck.
That pattern has a name in behavioural science: present bias. It means we systematically overvalue what's in front of us right now — the dopamine hit of a new tip, the novelty of trying something once — and undervalue the slower, duller process of actual skill-building. We're not lazy. We're wired this way. A single impressive demo feels like progress. Returning tomorrow, doing it again, slightly differently, feeling the gears turn — that doesn't feel like anything much at the time. But that's the thing that works.
So I built around that instead of against it.
The 15 minutes is not arbitrary. It's the number I kept landing on after thinking hard about what a busy professional can actually protect in a day without it feeling like a sacrifice. Not five minutes, which is too shallow to do real thinking. Not an hour, which requires a calendar block and a kind of commitment most people won't sustain past week two. Fifteen minutes sits in a particular sweet spot: long enough to actually try something, short enough that you won't negotiate with yourself out of it.
The workout format exists for a similar reason. Each session has a clear start, a task, and a finish. That's a habit loop — cue, routine, reward. The cue is the same time, the same place, the same format. The routine is the prompt exercise. The reward is the small, real satisfaction of having made something — a better email, a clearer brief, a tricky question answered — in a window that would otherwise have been Teams messages and mild dread.
What I want to be honest about: the first few days are not thrilling. You will write a prompt, it won't do what you wanted, and the temptation is to decide that AI is overhyped or that you're doing it wrong. Neither is true. A prompt that fails is just information. It tells you something about how you framed the question, what you assumed the AI knew, or what you actually wanted — which it turns out you hadn't quite worked out yet. That friction is the learning. I've tried to build the programme so that a bad output is never a dead end; it's a variable to adjust.
The other thing that shaped the design: reps. Not tips.
Tips feel generous but they're essentially someone else's reps handed to you as a shortcut. They work once, in one context, for the person who did the work to figure them out. What you actually need is the underlying feel — the sense of when to push a prompt further, when to change tack, when the problem is in the question and not the answer. That only comes from doing it yourself, repeatedly, with different material, until a version of that intuition is yours.
That's the IKEA effect, really. We value what we build ourselves more than what we're handed. But there's something more functional underneath it: you remember how you got somewhere. You can recreate it. You can adapt it. The person who reads a prompt tip and uses it once can't do any of those things.
This is why The AI Habit runs daily. Not because daily feels like discipline — I'm actually suspicious of anything that bills itself as a discipline — but because frequency is how you build the feel. A musician who practices for seven hours on Sunday is worse than one who practices an hour a day. Same total time. Very different outcome.
I started writing this as a defence of the design, and I realise it's also a confession. I've made my fair share of the mistakes above — the tips I collected and forgot, the tools I tried once and filed away. The programme is what I wish had existed when I was trying to get genuinely useful at AI rather than just theoretically aware of it.
If that sounds like something worth trying, the first 14 days are free. Fifteen minutes a day. No grand commitment, no upfront cost — just enough time to find out whether it actually does anything for you.
https://harrysharman.com/projects/ai-habit/