There is a cottage industry of “prompt engineering” advice — collections of techniques, magic phrases, and structural formulas for getting better output from AI. Most of it is more complicated than it needs to be. The reason most people get mediocre AI output is not that they lack technique. It is that their request contained no context. The technique that consistently improves AI output has nothing to do with engineering — it is the same thing that has always made communication work: tell the other party who you are, what you are trying to do, and what you need.
A better AI prompt is not a cleverly structured sentence. It is a brief.
What a Generic Prompt Is Missing
Take a request that produces generic output: “Write me a cover letter for a marketing job.”
The problem with this request is not the phrasing. It is the absence of almost all the information a useful cover letter requires. AI does not know: who is applying (their background, experience, industry), which specific job (the role, the company, the requirements), what the applicant wants to convey (their main pitch, their differentiating experience, their specific reason for wanting this role), or what good looks like (length, tone, what sections matter).
With none of that information, AI produces the average cover letter for the average marketing applicant for the average marketing role. If you are average, this is fine. If you are specific — which everyone is — it will not apply.
The Four Things Every Useful AI Request Needs
You do not need to learn any special techniques to write better AI prompts. You need to include four things:
**1. Your context.** Who are you in relation to this task? Not your whole biography — the relevant background. For the cover letter: “I have seven years in B2B marketing, focused on content and demand generation for technical products. I’m applying to a Head of Content role at a fintech startup.”
**2. The specific goal.** What are you trying to accomplish? Not just the task — the outcome. “I want to emphasize that I have built content programs from scratch twice before, which is exactly what this role requires.”
**3. The constraints.** What is the situation limiting you to, or what must the output include? “The job description emphasizes SEO and thought leadership. My background is heavier on thought leadership than SEO — the letter should acknowledge this without apologizing for it.”
**4. What a useful output looks like.** Format, length, tone. “One page, three paragraphs, confident and direct — not deferential. No summary of my resume — they have the resume.”
That is the brief. It is not a formula or a technique. It is the information a thoughtful colleague would need to help you with this task. AI needs the same information.
Before Context vs. After Context
Most people add context after the first response comes back generic. They read the generic output and say “but I’m actually in this specific situation” and AI adjusts. This works, but it is inefficient: you are doing the briefing work in the editing phase rather than the preparation phase.
The shift that produces consistently better AI output is moving the context from after to before. Instead of asking and then correcting, brief first. Establish the context, the goal, the constraints, and the desired output format before the request. The first response is much closer to useful because AI had what it needed before it started rather than partway through.
This feels slower initially — you are writing more before you see anything back. In practice it is faster, because you are not in a correction cycle.
The Role Line: The Single Most Efficient Improvement
If you only add one thing to your AI requests, add a role line at the beginning. “You are helping a [description of who you are] with [description of what you’re doing].” This single line moves AI from calibrating to the generic case to calibrating to your specific case.
Examples:
– “You are helping a first-year teacher at a Title I school prepare a parent communication.”
– “You are helping a solo founder with no marketing background write a LinkedIn post about a product launch.”
– “You are helping a midcareer professional write a resignation letter that preserves the relationship with their employer.”
Each of these produces dramatically different output for the same type of request — because AI now knows who it is writing for. The role line is not a trick. It is context, compressed into one sentence.
Better Prompts Come From Better Thinking, Not Better Technique
The most important thing a prompt can contain is clarity about what you are actually trying to accomplish — not what you want AI to do, but what you want to be true when it is done. That clarity comes from thinking before prompting. The brief is the artifact of that thinking — the thing you write before the request that contains what you know about your situation and what you need. Write the brief. The prompt writes itself.
Briefing Fox is built on this principle — helping you build the brief before any AI task, so the output is specific to you from the first response. Try it free at www.briefingfox.com.
Before Your Next AI Request
Before typing any AI request, take sixty seconds to write: who you are in relation to this task, what you are specifically trying to accomplish, and what a useful output looks like. That minute of preparation is the brief. The response that comes back is not generic.
Try Briefing Fox free at www.briefingfox.com.
Context about who you are in relation to the task. This single input moves AI from calibrating to the average person asking the question to calibrating to your specific situation. A one-sentence role description — “You are helping a first-generation college student write a cover letter for an entry-level marketing role” — changes the output more than any other addition.
Think of it as briefing a smart colleague rather than writing a command for a machine. Tell them who you are, what you’re trying to do, what constraints you’re working within, and what a useful output looks like. That’s a brief — no special technique required.
A role line is an opening sentence that establishes who the AI is helping and what context they’re in. “You are helping a senior product manager at a B2B SaaS company write a stakeholder update” tells AI who to calibrate to before any task is described. It’s the single most efficient improvement to any prompt.
Long enough to include the context, goal, and output specification — short enough to write in under two minutes. A well-briefed prompt is typically two to five sentences for most tasks. Longer is not better; more specific is better. Every sentence should be doing work that changes what AI produces.