People02 Oct 2026

It guessed because you didn't say

By Alan Goh · 2 min read

Before you ask

Fill the blanks you care about

  • Who is it for?
  • What is it for?
  • What must it include?
  • What must it avoid?
  • How should it sound?
Short answer

When AI gives you a generic answer, it has usually filled the details you left out with its best guess. Telling it the details you care about changes the answer.

Many people who have tried AI have had the same experience. You ask it for something, such as a reply to a customer, a summary of a report or a message for staff, and what comes back is generic. A bit off. Not quite something you’d send.

It’s easy to conclude that AI isn’t suited to your business. More often, it simply wasn’t told enough.

Every blank gets filled

AI fills in every detail you leave out, and it fills it with its best guess: the most likely answer across everyone who has ever asked something similar.

That is why the answer feels generic. It is the average answer to the average version of your question.

Draw a house

In my sessions I ask the room to draw a house, with ten seconds and no other instructions. Thirty people produce thirty different houses: tall, flat, with a chimney, without.

Then I ask again. A single-storey house, white walls, a red triangular roof, one door in the centre, a window on each side, a mango tree on the right.

The drawings come back nearly identical. Same people, same pens. The difference was how much was left to guess.

Spot the blanks

Take three ordinary requests.

“Reply to this angry customer.” Apologise? Offer something, and what? What actually happened? Firm or warm?

“Summarise this report.” How long? For whom? Which part matters most?

“Write a thank-you message.” To whom? For what? How formal?

Each question you leave open, AI answers for you.

Fill the ones you care about

You don’t need to fill every blank. That would take longer than writing it yourself.

Fill the ones that matter: who it is for, what it is for, what it must include, what it should avoid, and how it should sound. Leave the rest to its judgement.

One habit helps. Before you send a request, read it as a new member of staff would, someone with no background at all. What would they need to ask you? Those are the blanks.

The test

When an answer comes back off the mark, ask: what did I leave it to guess?

The AI didn’t get smarter the second time. Your instruction did.

In your department

  • Sales & Marketing: Before asking for a customer reply or a follow-up email, say who the customer is, what happened, what you can offer and how warm it should sound.
  • HR & Admin: A staff announcement comes back generic when AI has to guess the audience, the reason and the tone. Tell it all three.
  • Strategy & Planning: When you ask for a summary, say who it is for and which decision it supports, and AI will pick out the figures that matter.

Want to try this with your team?

Tell me what your team is working on, and I'll suggest where to start. I read and reply to every message myself.

More notes

  • 19 Sep 2026

    Hand AI the boring task first

    The best first task for AI is a routine one: it comes back every week, you could write its steps down, and you can check the answer. The impressive one can come later.

    Read it →
  • 09 Sep 2026

    Five skills for getting good results from AI

    Getting good results from AI comes down to five skills: break the task down, give it the context, set the limits, refine the draft, and judge the result. When an answer disappoints, the missing skill tells you what to fix.

    Read it →

All the notes →