Expert Bit: In the AI age, the highest-return investment isn’t a new tool. It’s a small stack of skills that makes every tool you touch work better — and it costs less than a month of most subscriptions.
The problem with learning AI
Most people learn AI backwards.
They start with the tool — ChatGPT today, Claude tomorrow, a new agent next week — and hope the tool will make them smarter. It won’t. A powerful tool in untrained hands just produces confident mediocrity faster.
The people getting real leverage from AI aren’t collecting tools. They’re building a skill stack: a few durable skills that compound. Each one makes the others stronger, and together they make any AI tool more useful.
This is that stack. Three skills. Under $100 total. In the right order.
- Clear writing — so you can think clearly, and tell an AI exactly what you want.
- Prompting / directing AI — so you can turn that clarity into useful output.
- Systems thinking — so you can turn one-off outputs into a repeatable system that works while you’re not watching.
Learn them in that order. Most people try to start at #2. That’s why their prompts don’t work.
Skill 1: Clear Writing — The Foundation
What it is: Saying what you mean, in as few words as needed, in the order a reader needs them.
Why it comes first: AI doesn’t read your mind. It reads your words. If your request is vague — “make this better,” “write something about marketing” — the AI has to guess. And it will guess wrong in a very polished way.
Clear writing is also clear thinking. If you can’t explain a problem in three sentences, you don’t understand it yet. Writing it down forces the understanding.
What to practise:
- The 3-sentence brief. Before you ask an AI for anything, write: (1) What I’m trying to do, (2) Who it’s for, (3) What good looks like. If you can’t fill those three lines, you’re not ready to prompt yet.
- Cut 30%. Take something you wrote — an email, a post, a plan — and cut a third of the words without losing meaning. Do it weekly. Your writing gets sharper, and so do your prompts.
- Name the outcome, not the task. Don’t write “write a blog post.” Write “write a 600-word post that convinces a busy small-business owner to try one automation this week.” The second version tells the AI what success looks like.
How to learn it for under $20:
Start free. Read the best short writing on the internet for 15 minutes a day — not to copy style, but to notice structure: short sentences, one idea per paragraph, a clear point at the top.
Then spend once, if you want a book to keep on your desk: a classic on clear writing, second-hand or ebook, usually $10–$20. You don’t need a course for this skill. You need reps. Write one clear brief a day for 30 days. That’s the course.
How you’ll know it’s working: Your AI outputs need fewer rewrites. Your emails get shorter replies — because people actually understand them. Your ideas stop sounding fuzzy, even to you.
Skill 2: Prompting / Directing AI — The Multiplier
What it is: Not “prompt engineering” as a magic trick. It’s directing: giving an AI a role, a task, context, constraints, and a way to check its own work — then judging the result like an editor, not a fan.
Why it comes second: Once you can write clearly, prompting becomes easy. A good prompt is just a good brief, plus a few AI-specific habits.
The 5-part prompt that works for almost everything:
- Role: “You are a [specific expert] helping a [specific person].” Not “you are an AI.” Be specific: “You are a pricing consultant helping a freelance designer.”
- Task: One verb, one outcome. “Draft,” “compare,” “rewrite,” “list” — then what done looks like.
- Context: The facts the AI needs and doesn’t have. Your audience, your budget, your deadline, what you’ve already tried.
- Constraints: Length, tone, format, what to avoid. “Under 300 words. No jargon. No hype. Bullet list.”
- Check: “Before you answer, list what you’re unsure about. After you answer, give me three ways this could be wrong.” That one habit catches more errors than any other.
What to practise:
- One prompt, three versions. Take a real task and write the prompt three ways: vague, clear, and clear + constraints. Compare the outputs. You’ll never write a vague prompt again once you’ve seen the difference side by side.
- The follow-up, not the re-prompt. Instead of starting over when output is off, say: “Keep the structure, but make the second half more specific, with one example.” Editing in conversation is faster than re-rolling.
- Build a prompt library. Save the 10 prompts that actually worked for you — in a note, a doc, anywhere. Reuse beats reinvention. Most of your work is the same 10 tasks in different clothes.
How to learn it for under $50:
You don’t need a $500 AI course. Use the free tier of a major AI tool daily — that’s your lab. Spend your money, if you spend it, on one focused short course or a month of a paid AI plan so you can practise with the better model and save your best prompts. Budget around $20–$50 for that month. Cancel after, keep the library.
The real cost is 20 minutes a day for 30 days, on real work — not toy exercises. Prompt on a real email, a real plan, a real post. Toy prompts teach toy skills.
How you’ll know it’s working: You stop copy-pasting the first answer. You start getting usable first drafts, and you can tell in 10 seconds whether an answer is right — because you knew what “right” looked like before you asked.
Skill 3: Systems Thinking — The Compounder
What it is: Seeing the workflow, not just the task. Turning “I used AI to write a post” into “I have a system that turns an idea into a drafted, checked, and scheduled post every Tuesday.”
Why it comes last: You can’t systematise what you can’t do once. First you learn to write clearly, then to direct AI well. Only then does it make sense to ask: how do I make this repeatable?
What it looks like in practice:
A one-off is: you prompt an AI to write a product description.
A system is: idea → 3-sentence brief → prompt from your library → draft → your 2-minute check (facts, tone, claims) → save to a folder → schedule. Same steps, every time, with a checklist.
Systems thinking is mostly three questions, asked about any repeated task:
- What triggers this? A day? A request? A new idea?
- What are the steps, in order? Write them down. If you can’t list them, you don’t have a system — you have a habit.
- Where does it break? The step you skip, the check you forget, the file you can’t find. Fix that one step first.
What to practise:
- Map one workflow this week. Pick something you do at least twice a week — invoicing, posting, planning, replying to the same question. Draw it as 5 boxes on paper. Then ask: which box could AI do, which box needs me, and which box is just waste?
- The 2-minute check. Every AI output gets the same check before you use it: Are the facts right? Is the tone right? Is there a claim I can’t defend? Systems fail silently when the check is skipped.
- Automate the boring middle, not the judgement. Let AI draft, summarise, and organise. Keep the decisions — what to publish, what to charge, what to promise — for yourself. That’s the part that compounds your reputation.
How to learn it for under $30:
This skill is mostly free. Read one short guide to workflows or checklists, watch one practical walkthrough of someone automating a real small-business task, then build one system for yourself. If you want a tool, spend $0–$30 on a month of a simple automation or notes app you actually use — not three tools you trial and abandon.
The test isn’t how fancy the system is. It’s whether you still use it in 30 days.
How you’ll know it’s working: You stop reinventing the same task. You have a folder, a checklist, and a prompt library — and a Tuesday that runs itself a little more than last Tuesday did.
The $100 Budget — A Sample Split
Prices change, and free tiers are enough to start. This is a sample, not a shopping list:
- Clear writing: $0–$20 — free reading + one classic writing book, second-hand or ebook.
- Prompting / directing AI: $0–$50 — free tier daily, plus one month of a paid AI plan or one short focused course to practise properly.
- Systems thinking: $0–$30 — free guides + one month of a notes/automation tool you actually use.
Total: $0–$100. The money is the smallest part. The 20 minutes a day for 30 days is the real investment — and it’s the part no tool can buy for you.
Spend in order. Don’t buy a systems tool before you can write a clear brief. Don’t buy an AI course before you’ve written 30 briefs by hand.
Your 30-Day Plan
Week 1 — Write clearly. One 3-sentence brief a day, on real work. Cut one thing you wrote by 30%.
Week 2 — Direct AI. Use the 5-part prompt on real tasks. Save every prompt that worked. Build your library to 10.
Week 3 — Map one system. Pick one repeated task. Draw the 5 boxes. Add the AI step and the 2-minute check.
Week 4 — Run it, then fix it. Run the system three times. Note where it breaks. Fix one break. That’s iteration — and it’s how a stack compounds.
After 30 days you won’t just “know AI.” You’ll have a way of working: clear briefs, reusable prompts, and one system that saves you time every week. Add a second system next month. Then a third.
That’s compounding. Small skills, used daily, getting more valuable the longer you use them.
The Expert Bit to take with you
Don’t chase the newest AI tool. Build the stack that makes every tool work better:
Write clearly. Direct AI well. Systematise what works.
Under $100. Thirty days. One skill at a time, in that order.
