Improving Results

Even a well-written prompt sometimes comes back close but not quite right. That's normal, and it's not a reason to start over — it's a reason to refine, the same way you'd hand a colleague notes on a first draft instead of redoing the work yourself.

A few techniques worth keeping handy

• Say exactly what's wrong. Instead of "make it better," try "this is too formal — make it sound like a quick note from a colleague," or "this is too long, cut it to 3 sentences."
• Ask for options. "Give me 3 versions of this headline, ranging from safe to bold" tends to surface something better than any single answer would.
• Fill in what's missing. A generic-sounding answer is almost always a missing-context problem — go back and add whatever detail you left out the first time.
• Have it critique its own draft. "What's the weakest part of this response, and how would you improve it?" often produces a noticeably sharper second attempt.

Treat every AI conversation as back-and-forth, not one-shot. The first response is a starting point, not a verdict on whether AI "worked" for the task. People who feel like AI "doesn't get it" are, more often than not, stopping after one try instead of refining — and refining is where most of the actual value shows up.