Your first twenty minutes.

Before anything else: forget the idea that you need to learn AI. You don’t. You need to get one small thing working, feel the click, and go from there. Everything below is aimed at that single outcome.

1. Pick one and stop shopping

Open any of the well-known assistants. Any of them. The comparison arguments you have seen online are real but they matter at the far edges of difficulty, not for what you are about to do.

The mistake almost everyone makes at this step is spending three days researching which to use and never getting to step two. Pick the one whose name you thought of first. You can change later and nothing is lost.

2. Choose something you already have to do

Not something impressive. Something tedious. The best first task is one you do repeatedly, already know the right answer to, and quietly resent.

• An email you have written some version of forty times
• A long document you need the gist of
• A messy list that needs organising
• A form letter, a description, a summary, a first draft of anything

Choosing a task you can already judge is the whole trick. You will be able to tell instantly whether the result is good, and that tells you far more than an unfamiliar task would.

3. Explain it like you would to a new assistant

Here is the entire skill, and it is not a trick: say what it’s for, who it’s for, and what you already know.

Compare these two.

“Write an email about the invoice.”

Versus:

“Write a short, friendly email to a long-standing client whose invoice is three weeks overdue. I want to keep the relationship warm — this is unusual for them and something has probably gone wrong on their end. Two short paragraphs, no guilt-tripping.”

Same tool. Wildly different result. You did not learn a technique; you just gave context you already had in your head.

4. Say what’s wrong with the first answer

The first answer is a starting point, not a verdict. Treat it exactly as you would a first draft from a capable person who does not yet know your preferences.

Useful things to say next: too formal. Shorter. You invented a detail I didn’t give you. Keep the second paragraph, redo the first. Less enthusiastic.

Most people stop after one round and conclude the tool is mediocre. The second and third rounds are where the value actually lives, and they cost you ten seconds each.

5. Check anything that matters

It will occasionally state something false in exactly the same confident tone it states something true. There is no tell. This is the one genuine hazard and the one rule worth memorising:

Use it for things you can verify, or things where roughly right is fine.

Dates, numbers, names, legal specifics, medical specifics, anything you would be embarrassed to be wrong about — check independently. Tone, structure, phrasing, brainstorming, summarising something you can see — low risk, high value.

6. Do the same task again tomorrow

This is the step everyone skips, and it is the one that decides whether any of this sticks.

Do not go looking for a second use case yet. Run the same task again tomorrow, and the day after. You will get faster, you will learn what context to give without thinking about it, and around the fourth or fifth time it stops feeling like a novelty and starts feeling like a tool you own.

That is the shift. It is much smaller and more boring than the marketing suggests, and it is the only part that matters.

What comes after

Once one thing is genuinely working, the honest map of what this is good and bad at will make far more sense than it would have this morning. After that, the guides go deeper on specific situations.

Common questions