How to use AI

Most people open ChatGPT, type one line, get something generic, and quietly conclude AI is overrated. The tool was fine. The input was one line with no context, so the answer was one line with no context.

A model only knows what is in front of it right now. Not your job, not your client, not what you decided yesterday. So the whole skill is deciding what to put in front of it, on purpose, before the question. Everything below is that one idea, made practical.

Do this once and you will feel the difference immediately. Before your real question, paste four short lines about yourself.

Then ask your actual question underneath. The same four lines work in ChatGPT, Claude and Gemini, and they will still work next year. Write them once, keep them in a note, paste them at the top of every chat.

A generic tool answering a generic question gives a generic answer. The same tool, told who you are, answers like it was built for you.

People lose weeks choosing. Do not. Pick one of the three big ones and start. They are all good enough, and switching later costs nothing because the notes you wrote are plain text that works in all of them.

The free tier of any of them is enough to learn on. Pay only once you hit a wall you can name.

Almost every "AI does not work for me" story is one of these.

• One line, no context.

  The single biggest cause. Fixed by the four lines above.

• Trusting it on facts.

  A model sounds equally confident when it knows and when it is guessing. It will invent a date, a source, a quote, in the same calm voice. Use it freely for shaping language. Check anything you are about to act on: names, numbers, citations.

• Expecting it to remember.

  Every new chat starts from zero. It does not recall yesterday unless you tell it again. That is not a bug you can fix with better wording, it is how the tool works, and the four-line note is how you work around it.

The same request, phrased two ways, gives two different answers. The pattern that works: say who it is for, give the raw material, say what shape you want back.

The second one is not longer because longer is better. It is longer because it carries the context the model cannot guess.

Once the four-line note becomes a habit, the next step is obvious. Instead of retyping it, you keep it in a file. Then you add a second file for the things it should remember about your work, and a third for the steps you repeat. Point your AI at those files at the start of a session, and it stops starting from zero.

That is an AI operating system, and it is nothing more exotic than a few plain text files you own. You still rent the model. You stop renting the context. If the four-line note made a difference, the full idea is the same move, repeated.

Write your four lines. Paste them into your next question. That is the whole first lesson.

Free starter files, including a note you can copy and edit, are on /resources. Everything here is free and needs no signup. If you want the next guide when it goes up, there is a waiting list, and that is the only thing here that asks for an email.