Prompting that works — AI Fundamentals for Business (module 2 of 6)

Last module: AI predicts plausible next words. It’s great at language work — but it knows nothing about your business until you tell it.

Talk to it like a smart new hire

Here’s the mental model that fixes most bad results: the AI is a smart new employee on day one. Capable — but it knows nothing about your business until you tell it.

You’d never hand a new hire a sticky note that says “write a job post.” You’d give them a brief. Prompting is that brief. Better brief in, better work out.

The four parts of a good brief

Contextwho you are

Taskwhat you want

Constraintsthe facts and limits

Formatwhat done looks like

A useful first draft

Four short pieces make one strong prompt — same brief you'd give a new hire.

See the difference, live

Same request, two briefs. Run both and compare what comes back — this is the whole lesson in one screen:

Weak prompt — vague in, generic out

Write a job post for a technician.

Strong prompt — same effort to type, completely different result

You're helping a 12-person HVAC company in Rapid City, South Dakota hire.

Write a job post for a service technician:
- 4+ years residential/commercial HVAC experience
- On-call rotation every third week
- $28–34/hr depending on experience, plus a company truck
- Tone: friendly and direct, like a small company that treats people well
- Under 300 words, ending with how to apply (email resume to jobs@ourcompany.com)

Iterate like a manager

The first draft is a starting point, not a verdict. The model remembers the conversation. So do what you’d do with a person: give feedback and ask for another pass.

“Good — but warmer, and half the length.”
“Drop the third paragraph, and don’t call us ‘dynamic.’”
“Give me three versions: safe, bold, and playful.”

Two rounds of feedback beats ten minutes of crafting the perfect first prompt. (Try it: run a prompt on the previous step, then use Edit & run to give it feedback.)

Show, don't tell

Adjectives like “professional” are weak signals. An example is a strong signal. Paste in something that sounds like you, and ask it to match:

Tone matching

Here's an email I wrote that sounds like me:

"Hey Mark — good news, the parts landed early so we can get you on the schedule Thursday instead of next week. Sarah will call to confirm a time. Thanks for hanging in there. — Dave"

Match that voice exactly. Now write an email to a customer whose repair is delayed one week because a compressor is back-ordered. Offer them first slot when it arrives.