AI doesn't know your industry. Here's how to fix that.
The prompts that teach AI your industry's real vocabulary and point of view
You can tell immediately. The phrasing is almost right but the texture is wrong. Words that no one in your field actually uses. A confidence about things that any practitioner would hedge. Generic examples that could apply to literally any business.
The output isn’t bad exactly. It’s just hollow. And if you send it out, your readers will feel that hollowness too, even if they can’t name it.
The problem isn’t the model. It’s that you’re asking it to write about your industry without ever actually teaching it your industry. AI defaults to the most average version of any topic it knows. That’s fine for a first draft. It’s a disaster for anything client-facing.
These 9 prompts fix that. They’re in sequence for a reason.
What you’ll get:
A prompt that builds your industry voice profile in under 10 minutes
A way to get AI sounding like a practitioner, not a Wikipedia entry
Prompts for calibrating jargon, spotting false confidence and generating insider-credible content
A reusable “industry context block” you can paste into any future session
Advanced prompts for critique, contrarian takes and competitive positioning
Prompt 1: Build your industry voice profile
What it does: Creates a reference document that captures how people in your field actually talk, think and write.
When to use it: At the start of any new AI session where you’ll be producing industry-facing content.
The prompt:
I work in [INDUSTRY/FIELD]. Before we write anything, I need you to build a voice profile for this industry.
Ask me 8 questions about how my industry communicates. Cover: the specific words and phrases practitioners use vs outsiders, the topics that insiders never need explained, what signals that someone doesn’t actually know the field, what’s currently debated or contested, and what the common misconceptions are.
After I answer, summarise everything into a “voice profile” I can paste into future sessions. Keep it under 300 words.
How to use it:
Run the prompt and answer all 8 questions honestly, including the embarrassing ones about what your industry actually argues about
Save the 300-word summary somewhere accessible
Paste it at the top of every future AI session before asking for content
Example input: Industry/field: B2B SaaS sales
What you’ll get: A compact reference document covering your industry’s real vocabulary, common misconceptions and the signals that separate credible insiders from people who read one blog post.
Advanced note: This profile gets better if you’re specific about sub-niches. “Marketing” is useless. “Performance marketing for DTC brands with sub-$50 AOV” gives the model something to work with.
Prompt 2: The jargon calibration test
What it does: Identifies which terms AI is using correctly, which it’s using loosely and which it’s just making up.
When to use it: After any first draft that’ll go to a professional audience.
The prompt:
I’m going to paste a piece of content about [INDUSTRY/TOPIC]. Your job is to audit the language.
For every industry-specific term or phrase, tell me:
Whether it’s used correctly in context
Whether a practitioner would actually use it this way, or whether it sounds like an outsider
Any terms that are being used too broadly or imprecisely
Flag anything that would make a [JOB TITLE/ROLE IN YOUR INDUSTRY] raise an eyebrow. Don’t be polite about it.
[PASTE YOUR DRAFT HERE]
How to use it:
Produce your first draft using any method
Run it through this prompt before editing
Fix the flagged terms, then re-paste and re-audit until nothing gets flagged
Example input: Industry/topic: Healthcare revenue cycle management. Job title: revenue cycle director.
What you’ll get: A specific list of terms that are wrong, vague or used in ways that signal unfamiliarity with the field. Not a vibe check. Actual line-by-line flags.
Advanced note: Adding “be brutal” to this prompt makes a real difference. The model’s default is to soften criticism. Tell it not to.
Prompt 3: Teach it what insiders never say
What it does: Removes the phrases and framings that immediately mark content as written by someone outside the industry.
When to use it: When a draft passes a basic read but still feels slightly off to you.
The prompt:
I work in [INDUSTRY]. Below is a list of phrases, words or framings that insiders never actually use. People outside the industry use them constantly when writing about us.
[LIST 5-10 PHRASES YOUR INDUSTRY FINDS CRINGEWORTHY OR IMPRECISE]
Here’s a draft I want you to revise. Remove any of these patterns. Where you remove something, replace it with how a practitioner would actually say the same thing. If you’re not sure how a practitioner would say it, flag the sentence instead of guessing.
[PASTE YOUR DRAFT HERE]
How to use it:
Spend 5 minutes making your list before running this. It’s worth it.
Run the revision
Review the flagged sentences yourself. Those are the gaps in your own voice profile.
Example input: Industry: Commercial real estate. Phrases insiders don’t use: “property portfolio,” “lucrative investment opportunity,” “prime location,” “capitalise on market trends,” “the commercial real estate space.”
What you’ll get: A version of your draft with the outsider signals removed and replaced with how the industry actually speaks.
Advanced note: Your list of phrases is the most valuable thing you produce in this step. Save it. Add to it over time. It becomes a filter you run on every future AI session.
You’ve seen what these prompts can do with a few minutes of setup. The next six go deeper. They cover the prompts that get AI producing content that sounds like it came from someone with 10 years in your field, not someone who read a Wikipedia summary. Prompts 4 through 7 take you up a gear.