If you've used AI tools — ChatGPT, Claude, Gemini — you've probably noticed the results can feel a little... flat. Generic. Like the output was written for no one in particular. You asked for a bio and got something that sounds like every other bio. You asked for marketing copy and got something that could have been written for literally any business.
The frustration is real. And the fix is simpler than most people think.
The Problem: AI Without Context Is a Contractor Without a Brief
When you open a blank chat and type "write me a bio," the AI has almost nothing to work with. It doesn't know your industry, your tone, your audience, or what you actually do. So it defaults to the safest, most average version of a bio it's ever seen. It's not being lazy — it genuinely doesn't have enough information to do better.
Most people respond by making the prompt longer and more detailed. That helps, but it misses the bigger lever: giving the AI a role before you give it a task.
The Role + Context + Constraint Formula
Before you ask AI to do anything substantive, give it three things:
- Role: Who is it? What's its job title, area of expertise, and perspective?
- Context: What does it know about your business, audience, and goals?
- Constraint: What are the guardrails — tone, length, format, what to avoid?
This is what separates people who get generic output from people who get output they can actually use.
A Practical Example
Here's the difference in practice. Say you run a San Diego acupuncture clinic and you want help writing a Google Business Profile description.
Without a role:
"Write a Google Business Profile description for my acupuncture clinic."
You'll get something serviceable but forgettable. Probably mentions "holistic healing" and "experienced practitioners" at least twice.
With a role:
"You are a healthcare copywriter specializing in integrative medicine practices. You write for patients who are skeptical of alternative medicine but open to trying it — they want evidence, not energy talk. My clinic is called [Name], located in [Neighborhood], San Diego. We specialize in pain management, fertility support, and stress relief. Write a Google Business Profile description — 250 words max — that leads with outcomes, avoids jargon, and ends with a clear call to action."
That second prompt gets you something you might actually publish without editing. The AI knows who it is, who it's writing for, and what success looks like.
Building a Reusable System Prompt
The best use of the role formula is to build it once and reuse it. Most AI tools — ChatGPT, Claude, and others — let you set a "system prompt" or "custom instructions" that applies to every conversation.
For a small business owner, a good system prompt might look like this:
"You are a senior copywriter and brand strategist for [Business Name], a [type of business] based in San Diego. Our audience is [describe them]. Our tone is [describe it — e.g., direct, warm, not corporate]. We never use [list phrases to avoid]. When writing for us, always lead with the benefit to the customer, not the feature of our service."
Set that once, and every task you give the AI — emails, social posts, blog drafts, service descriptions — benefits from it automatically. You stop re-explaining yourself every time. The AI starts operating like someone who actually knows your business.
What This Looks Like Across Tasks
Once you have a good system prompt, the same Role + Context + Constraint approach applies to everything:
- Email drafts: "You're our client success manager. Write a follow-up email to a prospect who went quiet after our initial consultation. Warm but not desperate. Two short paragraphs max."
- Social content: "You're our social media voice — conversational, never salesy. Write three Instagram captions for a before/after of a recent website project. Each caption under 80 words."
- Service descriptions: "You're our web copywriter. Rewrite this service page section so it speaks to the client's pain first, then introduces the solution. No passive voice."
In each case, you're not writing a longer prompt — you're giving the AI a job before you give it work.
This is one small part of how we help clients put AI to work in their businesses. It's not magic — it's just the difference between treating AI like a search engine and treating it like a contractor you've actually briefed.