Enabled User · Module 03

Context Is Fuel

After this, you include only the context the AI needs to do the job — no more, no less — and you understand why relevant detail produces better responses.

Intro

In E-02 you learned to protect the parts of your work you want left alone. Now you look at the other side: what you put in, before the AI starts. The verb sets what it does. The constraint sets what it leaves alone. Context is what lets it do either of those things well.

The concept

The AI cannot see your screen. It does not know your industry, your audience, your history, or your constraints unless you tell it. A message that assumes the AI already knows those things will get a generic response. A message that front-loads the relevant facts gets a specific one.

Context is fuel for machines, not a burden. You are not over-explaining. You are not wasting its time. The AI has no time to waste. More relevant detail produces a better response — every time.

The word to watch is relevant. Context is not "everything you know about the situation." It is the specific facts that change what the AI produces.

Without context:

Write me a presentation about our Q3 results.

The AI will produce something. It will be generic, balanced, and structured for an imaginary audience. It will not know that your audience cares about revenue and nothing else, or that you missed your target, or that you need the tone to be honest rather than defensive.

With context:

Write a 10-slide outline for our Q3 results. The audience is the senior leadership team — they care about revenue and customer retention, not operational detail. Revenue was up 12% vs Q2 but missed the 15% target. Customer retention dropped from 88% to 82%. The bright spot was the new product launch in September. Tone: honest and solutions-oriented, not defensive. Each slide: a title and 3-4 bullet points.

Same verb. Same task. The second version gives the AI what it actually needs to produce something specific.

The load-bearing test

Before you add a piece of context, ask one question: if I removed this, would the AI have to guess?

If yes — the AI would have to invent or assume something — it is load-bearing. Keep it.

If no — the AI would produce the same response either way — it is decorative. Leave it out.

Common types of load-bearing context:

Audience — who will read or use this, and what they care about
Purpose — what this is for, not just what it is
Tone — how it should sound, especially if the default would be wrong
Key facts — numbers, names, decisions, constraints that are specific to your situation
What has already been tried — so the AI does not repeat it

You do not need all of these for every task. You need the ones that would change the response if they were missing.

The exercise

Part 1 — Context audit

Find the opening seed you wrote in B-10 (or any real message you have sent recently).

Read it through. For each piece of information in it, ask the load-bearing test question: would the AI have to guess if this were missing?

Mark each piece:

Load-bearing (keep it — the AI needs this)
Decorative (remove it — the AI will not use it)

Then check what is missing. What would the AI have to guess that you have not told it? What is the audience? What is the tone? What are the constraints specific to your situation?

Rewrite the message: remove the decorative context, add the missing load-bearing context.

Part 2 — Before and after

Send both versions — your original message and your audited version — to your AI tool. Keep the task identical. Only the context changes.

Compare the two responses. Look for:

Specificity — does one response address your actual situation rather than a generic version of it?
Assumptions — where did the AI have to invent something in the first version that it did not have to invent in the second?
Length — a more specific response is often shorter, because the AI is not filling in blanks

You are not looking for "better." You are looking for what the context actually changed.

Copy-Personalise-Use

Starter

I'm a [role] working on [specific task]. The audience is [who will read or use this]. The key constraint is [one load-bearing fact]. [Verb] [specific thing]. [Stopping condition].

How to edit this

The three slots before the verb are the context layer. Fill in only what the AI cannot infer.

[role] — what you do, in a few words. "Marketing coordinator", "secondary school teacher", "small business owner."
[specific task] — what you are working on right now, in one sentence.
[audience] — who this is for, and what they care about. If the audience is irrelevant to the task, leave this slot out.
[one load-bearing fact] — the single most important piece of context. If you can only include one thing the AI does not know, what would it be?
[verb] — see E-01 for the verb reference.
[specific thing] — what you want the verb applied to.
[stopping condition] — word count, number of items, format. From B-10.

If you find yourself writing more than three or four context sentences, run the load-bearing test again. Something in there is probably decorative.

What good looks like

The response addresses your actual situation — not a generic version of the task. You can see that it used the context you gave it. There is nothing in the response that you would need to undo because the AI had to guess.

If the response is still generic

The context is probably still too thin. Ask yourself: what would make this task different from anyone else's version of it? The answer to that question is usually the load-bearing context you are missing. Add it and try again.

You now know what to put in (context), what to protect (constraints), and which word sets the scope (verb). E-04 goes deeper on what to do when, despite all of that, a conversation goes sideways and correcting it is making it worse.