What 90% of people in Geneva get wrong with AI

Hello Geneva,

A few months ago, a friend called me.

"Salim, I tried ChatGPT. Honestly... I don't get the hype. It spits out generic stuff. It's no better than Google."

I asked him just one question:

"Show me your last prompt."

He sent me a screenshot. His prompt? "Write me a professional email."

And right there, I understood his problem.

It's the same problem 90% of people have with AI. And probably yours too.

In fact, since GenevIA launched, it's THE question I get the most. In your emails, your LinkedIn messages, after my talks... Hundreds of you have asked me the same thing: "How do I get better at prompting?"

Let me explain...

The restaurant analogy

Picture yourself walking into a restaurant. The waiter comes over. And you tell him:

"Food, please."

What happens next?

The waiter gives you a strange look. He might bring you... something. But it won't be what you actually wanted.

Now picture yourself saying:

"I'd like a rare steak, with really crispy fries, a green salad on the side, and no sauce, please."

Now you get exactly what you want.

AI works exactly the same way.

"Write me an email" = "Food, please"

Mediocre result guaranteed.

The lightbulb moment

Once I understood that, everything changed.

I started treating AI like a brilliant new colleague... who knows absolutely nothing about my situation.

And I discovered 5 simple rules that completely transform the results.

Rule #1: Surgical precision

Bad: "Write me an email"

Good: "Write a follow-up email to a B2B client who hasn't replied in 2 weeks. We'd discussed a CHF 5,000 consulting engagement. Professional but warm tone. 6 lines max. End with an open question."

See the difference? In the second case, the AI knows exactly what to do.

Rule #2: Brief it like a new colleague

When you hand a task to someone who just joined, you don't just say "do the report."

You say: "It's for the CFO. He has 5 minutes to read it. He only wants the key numbers and the recommendation. Not the details."

Do the same with AI. Give it context: who it's for, why, and in what setting.

Rule #3: Spell out the format

AI can't read your mind.

Want a table? Say so. Bullet points? Say so. Flowing text in 3 paragraphs? Say so.

"Present the result as..." should end half of your prompts.

Rule #4: Positive beats negative

Your brain (and the AI's) struggles with negations.

"Don't make it too long, don't be too formal, avoid jargon"

vs.

"Write concisely, in an approachable tone, with simple words"

The second one works 10x better. Say what you want, not what you don't want.

Rule #5: Iteration

Nobody gets a perfect result on the first try. Not even me.

The magic is in the iteration.

"Make it shorter." "Add a concrete example." "Change the tone, it's too formal." "Keep only the first 3 points."

It's a conversation, not a vending machine.

These 5 rules are the foundation. But I bet you're thinking: "OK Salim, that's all very nice... but what do I actually paste into ChatGPT tomorrow morning?"

Good question.

I spent the past week pulling together my best prompts. The ones I actually use. The ones that work.

The result: a PDF guide with 35 ready-to-use professional prompts.

Follow-up emails. LinkedIn messages. Meeting minutes. Sales proposals. Feedback. Personal organization...

Tested formulas. Copy and paste in 30 seconds. No thinking required.

This guide will soon be available for CHF 19.

And my friend?

I rewrote his prompt using these 5 rules.

His reaction: "Wait... this is the same AI?!"

Yes. The same AI. Just a better way of talking to it.

See you next week,

Salim

🧞 Your wish is my command.

I'd love to hear what you think, and feel free to tell me what you'd like to see in upcoming issues!