Module 1, Lesson 1
In the next 15 minutes, you're going to use AI to produce something real for your work — then mark the places where it needs your professional judgment before it's ready to send.
By the end, you'll have two things: an AI-generated draft and your edited version with changes visible. That's the deliverable for this lesson — something you actually made, not something you read about.
If you're a Doer: Skip straight to "Your Turn" below. Come back to the walkthrough if you get stuck.
Pick one:
Don't overthink this. Pick the one closest to something on your to-do list right now.
Here's what the exercise looks like in practice. Sarah, an office manager at a 30-person accounting firm, needs to follow up with a software vendor who missed a demo appointment.
She opens ChatGPT (free tier) and types:
You are a professional office manager at an accounting firm. I need to write a follow-up email to a software vendor who missed our scheduled demo last Tuesday. The tone should be firm but professional — we want to reschedule, not burn the bridge. Keep it under 150 words.
30 seconds later, she has a draft. It's well-structured, grammatically perfect, and mostly usable. But she notices three things:
Sarah edits all three: fixes the date, adjusts the tone to match her voice, and adds the billing module requirement. Total time: about 4 minutes of editing on a 30-second draft.
What just happened: The AI produced a solid first draft. Sarah's professional judgment made it accurate and ready to send. Neither could have done the other's job as efficiently alone.
You are a [your role]. I need to [your task — be specific about what you need]. The [format/tone/length] should be [your requirements]. Make sure to [any constraints specific to your situation].
This structure — role, task, format requirements, and constraints — is the four-element prompt framework. You're using a simplified version here; Module 3 teaches you to construct it deliberately for any task. For now, just include what you'd tell a competent colleague if you were handing them this task.
Your edited version should be visibly different from the original in at least 2 places. If it isn't — if the AI output was perfect on the first try — you probably picked a task that's too simple, or you're not reading critically enough yet. That's fine. The next lesson explains why that critical eye matters so much.
If the AI output was completely unusable — gibberish, wrong topic, nonsensical — you probably gave it too little context. That's also fine. Module 3 will teach you exactly how to fix that.
You just produced a real work artifact with AI. You also discovered that the output needed your expertise. Next, let's understand WHY the AI gets some things right and other things wrong — the answer is simpler than you think, and it changes how you'll use AI from here on.