Give AI enough context to do the job
See how a vague request changes when you add the situation, the desired result, and the boundaries. The lesson turns prompting from a talent contest into a small specification exercise.
THE QUESTION THIS PAGE ANSWERS
ANSWER FIRSTWhat is the key idea behind “Give AI enough context to do the job”?
See how a vague request changes when you add the situation, the desired result, and the boundaries. The lesson turns prompting from a talent contest into a small specification exercise.
A prompt is a handoff, not a magic phrase. The model cannot use constraints you kept in your head. Put the audience, source material, format, and definition of done where both the model and a reviewer can see them.
Rewrite one vague request with context, task, constraints, and a finish line.
Adding adjectives like “expert” while leaving the actual decision unresolved.
Imagine a new colleague joins your company: top-school graduate, knows every field, executes brilliantly. But today is his first day on the job — he doesn't know you, doesn't know what you're responsible for, doesn't know who this thing is for or what it will be used for.
You walk over, drop a "write me a leave request", and walk away. What can he possibly produce? Only something invented from imagination. It's not that he isn't smart — it's that you told him nothing.
AI is that new colleague. And more extreme still: every new conversation is his first day on the job, all over again.
Scenario: you want to sell a used bicycle in your neighborhood group chat. Starting from the laziest possible ask, light up the four blocks of information below one by one and watch the AI's output improve step by step.
Who I am, in what situation
Your role, the occasion, who will read it, what it's for. This is the information the new colleague is missing most.
What exactly I want
The task + key details. The more concrete the details, the less room it has to make things up.
What counts as done
Tone, length, format, what to avoid. Give it acceptance criteria and it lands in one pass.
How “First, a thought experiment” becomes executable
“Imagine a new colleague joins your company: top-school graduate, knows every field, executes brilliantly.” is not about a magic phrase. It is about giving the model enough information to know who the work is for, what must be done, and what counts as acceptable.
Background sets direction; constraints set the boundary
“Scenario: you want to sell a used bicycle in your neighborhood group chat .” shows why a useful request separates the task, audience, source material, output format, and constraints. Without background, the model guesses. Without acceptance criteria, fluent text is not evidence that the task is complete.
- AI = a brilliant new colleague who doesn't know you — and every new conversation is his first day again
- Mediocre answers usually mean not enough information yet : give one unit of info, get one unit of answer
- The formula: context + request + constraints — who I am, what I want, what counts as done
More words do not guarantee a better result
Turn “Tone, length, format, what to avoid.” into a small experiment: change only one of background, requirements, or constraints while keeping the rest fixed, then observe which layer actually changes the output.
From “First, a thought experiment” to “Same task · Three ways of asking · Three results”
“First, a thought experiment” grounds the problem in “Imagine a new colleague joins your company: top-school graduate, knows every field, executes brilliantly. But today is his first day on the job — he doesn't know you, doesn't know what you're responsible for, d…”. “Same task · Three ways of asking · Three results” then moves it toward “Level 1 · Toss one line "Write me a leave request" Level 2 · Give some info Said how long and why Level 3 · Brief it fully Context + request + constraints You say to the AI The AI's answer”. Together, they show that the lesson is not just a conclusion to remember, but a claim with conditions.
Carry the judgment into the next situation
Build a request layer by layer: task and audience first, material and output rules next, constraints and acceptance checks last. Change one layer at a time so you know what actually helped.
- “First, a thought experiment”: Imagine a new colleague joins your company: top-school graduate, knows every field, executes brilliantly. But today is his first day on the job — he doesn't know you, doesn't know what you're responsible for, d…
- “Same task · Three ways of asking · Three results”: Level 1 · Toss one line "Write me a leave request" Level 2 · Give some info Said how long and why Level 3 · Brief it fully Context + request + constraints You say to the AI The AI's answer
- “The closing point”: Rambling at AI is free : thirty extra seconds of typing saves ten minutes of rework
The final “The closing point” brings the discussion to “Rambling at AI is free : thirty extra seconds of typing saves ten minutes of rework”. The useful thing to carry forward is knowing which judgments must be revisited when input, scale, or risk changes.
✅ What this lesson wants to share with you
- AI = a brilliant new colleague who doesn't know you — and every new conversation is his first day again
- Mediocre answers usually mean not enough information yet: give one unit of info, get one unit of answer
- The formula: context + request + constraints — who I am, what I want, what counts as done
- Rambling at AI is free: thirty extra seconds of typing saves ten minutes of rework
I turned one judgment from this article into a small experiment I could run today. Knowing what to observe next is more useful than simply remembering the conclusion.
After reading this, I first looked for the conditions behind the idea instead of copying the method into a project. That order made the later trade-offs much clearer.
When this judgment reaches real work, which constraint should be added first? I am curious which step matters most between reading and the first practical attempt.
No discussion on this article yet.