- Five Habits, Not a Course
Prompting for Business Users
You do not need a prompt engineering course. You need five habits that turn generic output into something worth using, and one rule about what never goes in the box.
SET EXPECTATIONS
The Skill Is Describing the Job, Not Knowing Magic Words
Most disappointing AI output comes from a request that would also have confused a capable new colleague. “Write a follow-up email” gives no indication of who it is to, what happened, what you want to happen next, or how your business sounds, so what comes back is generic, and the conclusion drawn is that the tool is not very good.
The good news for training is that this makes the skill teachable in about twenty minutes to anyone who can brief a colleague. It is not a technical skill and it does not need a specialist. The five habits below are what we cover with staff, and they hold across whichever tool you end up using.
We run that session inside Dallas offices, usually around a conference table, and we drive out from Plano for it. Prompting is taught faster in a room with the actual work on the screen than it is over a shared link.
THE FIVE HABITS
What Separates Useful Output From Generic
1
Say who it is for and what should happen next
“A referring physician who has not heard back in a week, and I want them to feel confident sending us the next one.” That single sentence does more than any formatting instruction.
2
Give it an example of your voice
Paste a past email you were happy with and ask for something in that style. Showing beats describing, and it is faster than explaining what “professional but warm” means to you.
3
State the constraints
Length, what must be included, what must not be said, and anything you cannot commit to. Constraints are usually where the professional judgment lives, and stating them is what makes the output usable rather than merely plausible.
4
Treat the first answer as a draft and push back
“Too formal.” “Shorter.” “You invented a deadline we never agreed to.” The second and third rounds are where the value is. People who accept or reject the first output are using a fraction of the tool.
5
Ask where it is uncertain
“What in this did you infer rather than take from what I gave you?” is the highest-value question a business user can ask, and almost nobody asks it. It surfaces exactly the parts that need checking.
The one rule that overrides all five
Habit two, giving it context, is genuinely the most effective of the five, and it is also how regulated data ends up somewhere it should not be. Context means the situation, not the patient’s record. Which tool you are signed into decides what you may paste, and that has to be settled before anyone is taught to prompt well. See the AI use policy template.
Do we need to hire a prompt engineer?
No. For a business of your size the skill belongs with the people who know the work, because the hard part is knowing what good output looks like, and that is domain knowledge, not a technical specialty.
Should we build a library of saved prompts?
For genuinely recurring tasks, yes, and at that point the better question is whether it should be a workflow rather than something a person retypes. A prompt used every day by six people is usually a sign of an automation waiting to be built.
Some of our staff find this uncomfortable. How do we handle that?
Start with a task they dislike rather than one they take pride in. Nobody is protective of reformatting a report; people are rightly protective of the work that makes them good at their job. Choosing the first task well resolves most of the discomfort without a conversation about it.
BK
Written and reviewed by Bob Kehr, President & Founder of Kehr Technologies, Plano City Council member, and Plano Chamber Small Business Person of the Year 2022, these five habits are what he covers with staff in the room, in about twenty minutes.
Last reviewed August 2026 · Questions we have not answered here? Call 214-444-3583. Kehr Technologies is based in Plano, Texas, and works with businesses throughout the Dallas area.