S-Tier Prompt Writing

The newer models are sharp. Depending on which one you're using, it might pause and ask you what chapter 7 actually covers before it starts. That's the smart move, and the good ones do it.

But plenty of the time, it just goes. Clean, confident, well-organized.

And it's never seen chapter 7. It doesn't have your textbook. So it's summarizing what a chapter 7 probably says, stitched together from every other textbook and every random page it ever read online.

Sometimes that's close. Sometimes your teacher's version is different. And you just studied the wrong thing.

You're trusting the internet over the book in your hands.

Sit with that for a second.

The page is open. You're reading it right now. And instead of showing the AI that page, you're asking it to answer from a fuzzy memory of a thousand other books and a pile of random stuff off the internet.

You picked the internet's version over the one you're literally studying from.

That's the real risk. Not that the AI is dumb. It's that you handed it the harder job and pointed it at the worse source, while the right source sat open in front of you.

When it fills a gap, it can fill it wrong.

Even the smart models do this. When they don't have your material and don't ask for it, they don't stop. They fill the gap with whatever sounds right.

A quote it half-remembers. A date that's close. A "fact" that fits the shape of the answer but isn't true.

People call this hallucination. It's not lying. It's confident guessing. And confident guessing is the hard kind to catch, because it reads exactly like the truth.

You don't want it guessing about your assignment. You want it working from the real thing.

Feed it the real thing.

The fix is almost stupidly simple. Give it the material.

Paste the actual chapter. Drop in your notes. Hand it your essay draft, your teacher's instructions, the exact passage you have to analyze.

Now it's not reaching into memory. It's reading what's in front of it. It's grounded.

Grounded means it can only work from what you gave it. The made-up quotes stop. The wrong dates stop. It's stuck to your reality, which is exactly where you want it.

This is not your situation.

Quick line so these two don't blur together.

Your situation is the circumstances around the request. Your material is the actual stuff it works on.

"I have a test tomorrow" is situation. The three pages of notes you're studying from is material.

One sets the scene. The other is the script.

What counts as material.

Anything real you can put in front of it.

• The source. The chapter, the article, the passage you have to read.
• Your own work. The essay draft you want feedback on. The half-finished lab report.
• The rules. The assignment sheet. The rubric your teacher actually grades with. The example they told you to follow.
• Your notes. The messy stuff from class that the AI has literally never seen.

If you're asking about something specific, and that something exists, paste it. Don't make it guess at a thing you could just show it.

See it in action.

Without it:

It writes a tidy summary. Generic causes. Maybe the same five your textbook lists. Maybe not. You can't tell, because it isn't actually looking at your book.

With it:

Now it's summarizing your book. Your teacher's framing. The exact points you'll be tested on. Not a generic version. The one that's on the test.

Same task. You just took away its ability to guess.

We already built this in.

Quick note, because it matters here.

Every prompt in the Vertech library is built to put your material to work. Paste your essay and the Essay Coach reads your essay. Drop in your notes and the Notes Organizer organizes yours. They're designed to ground themselves in what you give them, so you get help on your actual work instead of a generic version of it.

You can see the whole set in the Library.

It knows who it's for. It knows your situation. It's working from your material.
One piece left: telling it what you actually want out of all this. That's Your Goal.