The Context layer is the bottom of the Context Stack, and it answers three questions before you ask for any work: who is this business, who is the reader, and what is the situation. Two or three sentences. Skip it and every draft comes back sounding like a brochure for a company that does not exist.
Key takeaways
- Context is the shortest layer and the fastest to write. Two or three sentences covers it.
- It fixes the most common complaint about AI writing, which is that it sounds generic.
- Include the trade, the region, the customer and the moment in the job.
- Leave out mission statements, values and anything you would not say on the phone.
- Written once, it carries across every Stack you build afterwards.
What the layer holds
| Element | Example |
|---|---|
| What the business does | Roof repairs and re-roofs, residential |
| Where it operates | Sunshine Coast and northern Brisbane |
| Who the reader is | A homeowner who has just had someone on the roof |
| What they already know | They know the problem, not the fix or the price |
| The moment in the job | Post-inspection, pre-quote |
Five short facts. The last two matter more than people expect, because they set the tone of everything produced. A homeowner mid-quote gets a different document from a builder mid-tender.
How to write it
Say it out loud as though you were briefing a new employee in the ute on the way to a job. That is the register. "We do residential re-roofs on the Coast. This one is a homeowner in Buderim, she knows there is a leak, she has no idea what a re-roof costs, and I have just been up there."
Then write that down. It is already correct. Formalising it into corporate language makes it worse, because the draft then matches the formal register instead of yours.
What to leave out
Values. Mission. Years in business, unless it changes the answer. Awards. Anything that belongs on a website About page. The Context layer is operational, not promotional, and padding it with marketing language teaches the assistant to write marketing language back at you.
Common mistakes
Too broad. "We are a building company in Australia" produces drafts for every building company in Australia.
Too promotional. Feed it your brochure and you get brochure copy back, which is the most common way businesses end up with drafts that sound like everyone else.
Missing the reader. Describing your business but never the person reading the output leaves the assistant guessing at tone and detail level.
Rewritten every time. Context is stable. Write it once, keep it in the Stack, reuse it. Retyping it from memory each session is where variation creeps in.
Where it sits in the Stack
Context is layer one of five. On top of it go Task, Source, Rules and Output. Read the full Context Stack.
Build yours on a real job in the workshop: a full day at Avani Mooloolaba Beach Hotel, A$495 excluding GST, capped at 15 people. See the workshop, or enquire about onsite team training.
Frequently asked questions
How long should the Context layer be?
Two or three sentences. Longer does not improve output, and padding it with marketing language actively harms it.
Why does AI output sound generic?
Almost always a missing or vague Context layer combined with a missing Source layer. With no description of your business and none of your material, the average of everything is all there is to write from. See the full diagnostic.
Should you include your brand voice in the Context layer?
Describe the reader and the situation, and let the Output layer handle format. Voice comes through better from a real example of your own writing in the Source layer than from adjectives about tone.
Does the Context layer change between jobs?
Rarely. The business description stays. The reader and the moment in the job change, so keep those two as the editable lines.
Do you need a separate Context layer for each service?
One per distinct customer type is usually enough. A business quoting homeowners and tendering to builders needs two.
What if you are a sole trader?
Then say so. It changes how the draft handles scheduling, availability and who will be on site, which is exactly the kind of detail readers ask about.