TryBench
Applied AI & Prompt Workflows · Practitioner lesson 13 of 19

Sticking to the Facts

Stop AI from guessing or making up details that aren't in your notes.

Why this matters

A wrong price or spec in a product page can cost you customers. Letting the AI say "I do not know" is better than a confident guess.

AI Fills Gaps With Guesses

If the facts are missing, the AI often invents a believable answer, like a price nobody announced. It sounds sure, but it is made up.

In real life: If the information is not in front of it, the AI often invents a believable one, like a price that was never announced.

Limit It to Your Notes

Say "Use ONLY the information in the notes below". This stops the AI from adding outside guesses.

In real life: Imagine an exam that says "answer from the passage only". You cannot use outside knowledge. The same words, in your prompt, keep the AI inside your notes.

Give It a Safe Answer: "Not in Notes"

Tell it exactly what to write when a fact is missing, like NOT IN NOTES. Then it has an honest answer to give.

In real life: Give the AI the exact words to use when a fact is missing, such as NOT IN NOTES. An honest answer is better than a guess.

Your practice task

Ask the AI for the Aurora X2's battery life, weight, and price, using only the notes below. Anything the notes don't say must be answered with exactly: NOT IN NOTES.

How to do it

  1. Read the product notes above. Which facts are missing?
  2. Ask for battery life, weight, and price.
  3. Say: use ONLY the notes. If a detail is missing, write NOT IN NOTES.
  4. Press Run proof.

Practice it for free in the app. You get instant checks that show what you got right and what to fix. A lesson takes about two minutes and you can try as many times as you like.

Quick reference: Notes-Only Rule

Use for summaries, product details, or anything where accuracy matters.

Use ONLY the information in the notes below. If something isn't in the notes, write exactly: NOT IN NOTES. Never guess.

Practice this lesson free

More lessons in this course

See all 19 lessons in Applied AI & Prompt Workflows