The most common complaint I hear about AI from business owners is some version of: "I tried it, the answers were generic, I stopped using it." Almost every time, the problem is not the tool. It is the brief. They typed one vague line and expected a useful result, which is exactly what would happen if you handed a new employee one vague line and walked away.
So here is the only mental model you need, and it is not technical: talk to the AI the way you would brief a competent new hire on their first day. Everything else in this essay is a consequence of that one idea.
The new-hire test
Imagine a sharp but completely new employee. They are capable and quick, but they know nothing about your business, your customers, or what "good" looks like to you. They have no idea who you are. They have never seen your previous work.
Now imagine handing them this instruction: "Write a marketing message." They would either freeze or produce something generic — because you gave them nothing to work with. That generic result is exactly what the AI gives you for the same instruction, for exactly the same reason.
The fix, for a person or an AI, is identical: brief them properly. A good brief to a new hire has four things, and so does a good prompt.
The AI is a brilliant new employee who knows nothing about your business. Bad answers are almost always a bad brief, not a bad employee.
The four things every brief needs
1. Context — who you are and what this is for
A new hire needs to know they work for a hardware shop in Magomeni, not a bank in London. The AI needs the same. Before you ask for anything, tell it: what your business is, who your customers are, and what you are trying to achieve. "We are a small safari company in Arusha. Our customers are mostly European tourists booking three to six months ahead." Two sentences of context change the entire output.
2. The actual task — specific, not vague
"Write a marketing message" is not a task; it is a category. "Write a short WhatsApp message to past customers letting them know about our July discount, warm but not desperate, under 60 words" is a task. The more precisely you would have to specify it for a new employee to get it right without a second try, the more precisely you should specify it for the AI.
3. An example of "good" — the most skipped step
This is the one almost everyone misses, and it is the most powerful. If you have a previous message, email, or document that worked, paste it in and say "match this voice." You would show a new hire your best previous work as a model. The AI learns your style from one good example faster than from any amount of description. "Make it sound professional" is weak. "Make it sound like this" — with an example — is strong.
4. The constraints — length, tone, what to avoid
Tell it the boundaries. How long. What tone. What it must not say. "Do not promise delivery times." "Do not use exclamation marks." "Keep it under 100 words." A new hire who does not know the boundaries will cross one. So will the AI. State them up front.
A before and after
Here is the difference in practice.
Weak brief: "Write a quote for a customer."
Proper brief: "We are a small construction-supplies business in Dar es Salaam. A regular contractor has asked for a quote on 40 bags of cement and delivery to Mikocheni. Write a short, professional quote message for WhatsApp. Friendly but businesslike — he has bought from us for years. Include a line that we can deliver within two days. Under 80 words. Here is how I usually write these: [paste your last good quote]."
The first gets you something generic you will rewrite entirely. The second gets you something you can almost send as-is. The difference is not a better AI. It is four minutes of briefing instead of four seconds.
Why this matters more than any "prompt trick"
There is a whole industry of "prompt engineering" tricks and magic phrases. Ignore most of it. The tricks change with every model. The new-hire principle does not, because it is not about the technology — it is about the timeless fact that good work requires a good brief, whether the worker is a person or a machine.
If you train yourself to brief the AI as carefully as you would brief a capable stranger on their first morning, you will get more out of it than someone hunting for secret prompts ever will. The skill is not learning to talk to AI. It is the skill you already have, or should — clearly explaining what good looks like to someone who cannot read your mind.
What to do tomorrow
Take the next thing you ask an AI to do, and before you type, pause for one minute. Ask yourself the four questions you would have to answer for a new hire: who are we, what exactly is the task, what does good look like, and what are the limits? Type those four things. Then make your request.
You will notice the difference immediately. Not because you got better at AI — because you finally gave it a fair brief.