You ask AI to “write a follow-up email to a customer,” and back comes something bland and a little robotic — “I hope this email finds you well; I wanted to circle back regarding our recent conversation.” You read it, wince, and decide AI writing just sounds fake.
But look at what you gave it. “Write a follow-up email” is an order so vague that the only thing it can do is reach for the most generic version that fits. You didn’t get a bland answer because the AI is bland. You got a bland answer because you gave it a bland instruction.
This is the cheapest upgrade in all of AI, and it costs nothing: be specific. Vague in, vague out. Specific in, specific out. The AI isn’t a mind reader — every detail you leave out, it fills with the safest, most average guess. Your job is to leave out less.
Watch what specific does
Here’s the same task, done right: “Write a short follow-up to a customer, Dave, who got a quote last week for a $4,200 water-heater replacement and hasn’t replied. Friendly, not pushy — the way I’d talk to him at the counter. Three sentences. Remind him the quote’s good for 30 days and ask if he’s got any questions.”
Same AI. Completely different email — one you could actually send. The difference wasn’t a magic prompt or a better model. It was the who, the goal, the tone, the length, and the constraints. You handed it the details only you knew, and it did the typing.
Four ways to sharpen any ask
Name the specifics. Who’s it for, what’s the goal, how long, and the one thing it must include or leave out. The more real detail, the less guessing.
Show it what “good” looks like. Even faster than describing: paste an example. “Here’s a follow-up I sent that landed the job — match this style.” It learns your voice from one real sample quicker than from a paragraph of adjectives.
Say what TO do, not what to avoid. “Don’t be formal” backfires — it fixates on the word formal. Flip it to the target: “Write like you’d talk to a regular at the counter.” Aim it, don’t just warn it.
When you’re not sure how to ask — make it interview you. You don’t have to be articulate up front. Add: “Before you write this, ask me what you need to know.” Now it pulls the details out of your head one question at a time, instead of guessing. That single line turns a blank-page stare into a conversation.
The catch: specific isn’t the same as long
This trips people up. Being specific doesn’t mean writing a giant wall of instructions. A sharp two-sentence ask beats a rambling paragraph every time. You’re not writing more — you’re writing the right details, the handful that actually change the answer. And if you genuinely don’t know which details matter, don’t fake it — that’s exactly what move #4 is for. Let it ask you.
Why this is the upgrade that pays off everywhere
Here’s the part worth sitting with. Specificity isn’t a trick for better emails — it’s the skill underneath everything you’ll ever do with AI. A quick email today, a cleaned-up spreadsheet tomorrow, and eventually a real workflow: “every morning, pull the day’s jobs, draft the customer texts, flag the ones missing a deposit.” The thing that separates an automation you trust from one that quietly does the wrong thing is the very same thing that separates a good email from a bland one — clear, specific, unambiguous instructions.
So you’re not just learning to write better prompts. You’re learning the language you’ll use to hand off bigger and bigger jobs. Practice it on the small stuff now, and it compounds into the ability to delegate the real stuff later.
Carry this out
Stop typing vague and hoping. Type specific — or tell it to interview you. It’s the closest thing to a free upgrade you’ll find.
Reply and tell me: what’s a task where AI keeps handing you bland, generic results? Send me the prompt you’ve been using and I’ll show you the specific version.
Next week: how to get answers you can actually trust — by pointing AI at your own documents instead of the whole internet.
"}]