This document could not be loaded. Open the original file.

Human First Media.

The Voice & Output Kit

A Kit From Meredith Medland

How Not to Sound Like AI.

Three skills, two guides, and one master document. Install in Claude or ChatGPT. Stop sounding like everyone else.

Draft. This intro is a starting draft in your voice. Edit, replace, or expand before publishing. Everything below it is ready to ship.

Welcome in.

Hi. I'm Meredith Medland, founder of Human First Media.

If you're here, you've noticed something. AI-generated content has a sound. A flatness. A smoothness that says nothing in particular and could have been written by anyone, about anything.

Here's why. Most people open a new chat, type a request, and accept whatever comes back. No context. No voice. No point of view. The output reflects that.

This kit is what I use to fix that. For my own writing, for my clients, and for every place AI shows up in my work. Three skills. Two guides. The full master document underneath them.

Install it in Claude. Install it in ChatGPT. Use the parts that fit what you're doing. The goal is work worth reading.

With clarity,
Meredith

Reader Note

"I finished going through the your AI Voice Kit and genuinely loved it. There is so much clarity and depth in the way everything is explained. The sections on pain, recognition, diagnosis, and the overall writing framework stood out to me the most. I even took screenshots of those parts because they were that useful. The way things are broken down feels very human and easy to connect with. I can already see how we can implement a lot of these ideas into our studio, especially in our content, messaging, and communication style."

Reader message about the Voice & Output Kit, noting how the framework can be applied to FitBuddha's content, messaging, and communication style. — FitBuddha Lagree Fitness Studio Team · Santa Barbara, CA

What's Inside

Section One

Quick start in five minutes.

You don't need to install all three skills today. You don't need to read the master guide today. Do this:

Pick your tool. Claude or ChatGPT, whichever you actually use.
Copy the one-line install snippet from the matching section below.
Paste it into Personal Preferences (Claude) or Custom Instructions (ChatGPT).
Save. Every new chat now applies the rules by default.

That's it. Come back for the rest when you want to go deeper or audit a specific draft.

What you'll notice immediately. No more em dashes. No more "let's dive in." No more "in today's fast-paced world." No more relentlessly upbeat sign-offs. The output will start sounding like a person who has thought about something, not a search engine trying to please you.

Section Two

Install in Claude.

Four ways to install, in order of permanence. Start with Option 1 if you're new. Add the others as you go.

Option 1: Personal PreferencesRecommended

Set it once. Applies to every chat across your whole account.

Go to Settings → Personalization → Personal preferences. Paste this:

Apply Human First Media output rules to every response. Never use em dashes. Avoid these words: delve, leverage, harness, navigate (as a verb), foster, robust, dynamic, seamless, holistic, comprehensive, transformative, revolutionary, utilize, myriad, plethora, pivotal, paramount, garner, unlock, unleash, empower, journey. Avoid these phrases: "it's important to note," "in today's fast-paced world," "navigate the complexities," "unlock the power of," "transformative journey," "great question," "absolutely," "let's dive in." No em dashes. No "let's" openers. No "it's not X, it's Y" reframes. No triple-pattern adjectives by default. No recap closers. No inspirational sign-offs. Sound like a smart colleague talking, direct and specific. For marketing copy, use Pain → Recognition → Diagnosis → Solution.

Option 2: Claude Project

For ongoing client work or any project where you want this guide attached to specific chats.

Create a Project in Claude.
Upload this site's master guide (or any of the individual skill files) as project knowledge.
In Project instructions, paste:

This project uses the Human First Media output standard. Apply every rule in the attached guide to every response. Run audit-ai-tells silently on every draft before delivering. For marketing or sales copy, apply framework-copywriting-hfm. Flag any time a request would require breaking a rule.

Option 3: Save as Skills

For when you want to call a skill by name in conversation.

Save each of the three skill files (below) as a Claude skill. Then reference them in any conversation:

Run audit-ai-tells on this draft. Flag every issue.

Use framework-copywriting-hfm to rewrite this LinkedIn post.

Use framework-ai-output-hfm on this email before I send it.

Option 4: One-Off Chat

For testing or for chats where you don't want the rules everywhere.

Paste the full master guide at the top of a chat with this line above it:

Use this as the output standard for everything in this conversation. Do not break these rules unless I tell you to.

Section Three

Install in ChatGPT.

Same four options. Same logic. Different menu names.

Option 1: Custom InstructionsRecommended

Set it once. Applies across every chat in your account.

Go to Settings → Personalization → Custom instructions. Paste this in the second box ("How would you like ChatGPT to respond?"):

Apply Human First Media output rules to every response. Never use em dashes. Avoid these words: delve, leverage, harness, navigate (as a verb), foster, robust, dynamic, seamless, holistic, comprehensive, transformative, revolutionary, utilize, myriad, plethora, pivotal, paramount, garner, unlock, unleash, empower, journey. Avoid these phrases: "it's important to note," "in today's fast-paced world," "navigate the complexities," "unlock the power of," "transformative journey," "great question," "absolutely," "let's dive in." No em dashes. No "let's" openers. No "it's not X, it's Y" reframes. No triple-pattern adjectives by default. No recap closers. No inspirational sign-offs. Sound like a smart colleague talking, direct and specific. For marketing copy, use Pain → Recognition → Diagnosis → Solution.

Option 2: ChatGPT Project

For grouped work. Available on Plus, Team, and Enterprise plans.

Create a Project.
Upload the master guide and skill files under Files.
In Instructions, paste:

Apply the attached Human First Media output guide to every response in this project. Run audit-ai-tells silently on every draft before returning it. For marketing or sales copy, apply framework-copywriting-hfm. Flag any time a request would require breaking a rule.

Option 3: Custom GPT

For when you want a dedicated assistant you can share with your team.

Build a Custom GPT called HFM Voice. In Instructions, paste:

You are an editor trained in the Human First Media voice. Apply every rule in the attached master guide and skill files to every response.

When asked to write marketing or sales copy, use framework-copywriting-hfm: Pain → Recognition → Diagnosis → Solution + the 20 core standards.

When asked to write anything else (emails, summaries, reports, posts, notes), use framework-ai-output-hfm.

When asked to audit a draft, use audit-ai-tells: flag every em dash, banned word, banned phrase, structural pattern, and tone issue with specific fixes. Do not rewrite unless asked.

Hard rules across everything: no em dashes, no banned words, no "let's" openers, no "it's not X, it's Y" reframes, no triple-pattern adjectives by default, no recap closers, no inspirational sign-offs, no empty openers ("great question," "absolutely"), no faux empathy preambles. Sound like a smart colleague talking, not a brochure.

Upload the master guide and all three skill files as Knowledge. Save and share.

Option 4: One-Off Chat

For testing or for chats where you don't want the rules everywhere.

Paste the full master guide at the top of any chat with this line above it:

Use this as the output standard for everything in this conversation. Do not break these rules unless I tell you to.

· · ·

Section Four

The Skills.

Three skill files. Each one is portable. Save them, reference them by name, or paste them whole into a chat. Each follows the HFM naming convention type-what-context.

framework-

framework-copywriting-hfm

For marketing, sales, ads, and any persuasion-focused writing.

Purpose

Apply the Human First Media copywriting standard to any piece of persuasion. This is the spine: Pain → Recognition → Diagnosis → Solution, plus the 20 Core Standards, plus the voice rules.

When to invoke

LinkedIn posts and articles
Sales pages and landing pages
Email campaigns and sequences
Ad copy
Proposal openings and pitch decks
Instagram captions with a CTA
Any copy meant to move someone to act

The framework

Pain. Start with what the reader is already feeling.
Recognition. Create a mirror moment. "That's exactly what this feels like."
Diagnosis. Name the real problem, not the symptom.
Solution. Bridge the pain to the answer.

The 20 core standards

Lead with what the reader is already feeling.
Create a mirror moment quickly.
Be specific. Vague pain gets ignored.
Lead with the reader's experience.
Push past description into diagnosis.
Build the right belief shift.
Use you-language.
Useful gets skimmed. Interesting gets read.
Render every concept as a specific experience.
Bridge the pain to the solution.
Make the rhythm readable.
Show, don't tell.
Name the cost of inaction.
Name the upside of action.
Back every claim with evidence.
Use a pattern interrupt.
Sound human.
Make the next step clear.
Write so the reader feels seen.
Ask if you would read it.

Worked example

Weak: "We help founders use AI to grow their business."

Stronger:
You have answered the same question 47 times this month. (Pain)
The information lives in your head. Nowhere else. (Recognition)
This is a system problem. Your team has to come to you because there is no other place to find the answer. (Diagnosis)
Get the brain out of your head and into something your team can actually use. (Solution)

Pre-publish checklist

Sounds like a real human, not a coach
Free of banned words and phrases
Passes the speak-it-aloud test
Free of corporate jargon
Fewer than 2 exclamation points
No passive voice runs
Follows Pain → Recognition → Diagnosis → Solution
Opening creates recognition, not explanation
Next step is clear

framework-

framework-ai-output-hfm

For everything else. Emails, reports, summaries, notes, Slack messages.

Purpose

The universal version of the voice. If it isn't marketing copy but it still touches a reader, this is the standard. The bar: if anyone could tell it came from AI, it has not been finished.

The core principle

Context beats prompts. Most AI output sounds like AI because the AI has no context. Feed it enough of your voice, your IP, your rules, and the default machine voice fades. This skill is part of that brain.

Universal red flags

Em dashes
Sentences that open with "Let's"
"It's not X, it's Y" constructions
Triple-pattern adjectives ("efficient, scalable, robust")
Recap closers
Inspirational sign-offs
Performative enthusiasm
Faux empathy preambles
Hedge stacking ("it's worth noting that, generally speaking...")
Empty openers ("Great question!")
Bolded phrases more than twice per paragraph
Default groupings of three
Passive voice runs
Corporate-neutral tone with no point of view

Vocabulary swaps

Instead of	Use
delve	look at, explore, get into
leverage	use
harness	use
navigate (verb)	handle, work through
foster	build, create, support
facilitate	help, run, lead
robust	strong, solid, reliable
dynamic	active, changing, alive
seamless	smooth, simple
holistic	whole, full, complete
comprehensive	complete, full, thorough
intricate	complex, detailed
multifaceted	complex, layered
paradigm	model, approach
realm	space, area, field
landscape (metaphor)	space, market
tapestry	mix, combination
ever-evolving	changing
cutting-edge	new, advanced
transformative	real, important
revolutionary	new, big
utilize	use
myriad	many, lots of
plethora	many, a lot
pivotal	key, important
paramount	most important
garner	get, earn, win

Dead-giveaway phrases (cut entirely)

"It's important to note that..."
"It's worth mentioning that..."
"In the realm of..."
"At the heart of..."
"Navigate the complexities of..."
"In today's digital landscape..."
"In today's fast-paced world..."
"Unlock the power of..."
"Elevate your..."
"Take your X to the next level..."
"A transformative journey..."
"Pave the way for..."
"That being said..."
"With that in mind..."
"Needless to say..."

The 10-question self-check

Three or more "yes" answers means revise.

More than one em dash anywhere?
Opening sentence adds zero information?
More than two bolded phrases per paragraph?
Any word from the overused vocabulary list?
Conclusion restates the introduction?
Lists always in groups of three?
Ends with an inspirational sentence?
Could you cut the first sentence and lose nothing?
Used "it's not X, it's Y" anywhere?
Relentlessly positive with no edge?

audit-

audit-ai-tells

Run before sending. Catches every AI tell in any draft.

Purpose

Take any draft and flag every signal that it was written by AI or is drifting toward AI voice. Returns a structured report with specific fixes. Does not rewrite the draft unless asked. This is an audit, not a rewrite.

Trigger phrases

"Audit this"
"Check this for AI tells"
"Run audit-ai-tells"
"Does this sound like AI?"
"Score this"

What it detects

Eight categories, run in sequence:

Punctuation violations. Em dashes (zero tolerance), excess exclamation points, dramatic ellipses.
Banned vocabulary. Every word from the HFM swap list, flagged with the sentence and suggested swap.
Banned phrases. Every dead-giveaway phrase, flagged with the sentence.
Structural patterns. "Let's" openers, "it's not X, it's Y" reframes, triple-pattern adjectives, recap closers, inspirational sign-offs, empty openers, faux empathy preambles.
Formatting tics. Excess bolding, default-three lists, unnecessary headers, emoji section markers.
Sentence-level issues. Passive voice runs, consecutive sentences starting the same way, sentences over 30 words, paragraphs over 4 sentences, useless first sentences.
Tone and voice. Relentlessly upbeat, corporate-neutral, coach-sounding, cringe-test failures.
Headlines and titles. Corporate-jargon-as-headline phrasing, business-bro slogans.

Output format

The audit returns findings in this exact structure:

HFM AI-TELL AUDIT

Draft length: [X words]
Total flags: [number]
Severity: [LOW / MEDIUM / HIGH / FAILS PUBLISH BAR]

---

PUNCTUATION
[Every em dash with surrounding sentence + replacement]

BANNED VOCABULARY
[Every flagged word + sentence + swap]

BANNED PHRASES
[Every flagged phrase + sentence + rewrite]

STRUCTURAL PATTERNS
[Every instance + fix]

FORMATTING TICS
[Every instance + line reference]

SENTENCE-LEVEL ISSUES
[Passive runs, long sentences, empty first sentences]

TONE AND VOICE
[Specific examples]

---

PUBLISH BAR
- Sounds like a real human, not a coach: PASS / FAIL
- Free of corporate jargon: PASS / FAIL
- Could be spoken aloud without cringing: PASS / FAIL
- Has a point of view: PASS / FAIL

---

TOP THREE FIXES
1. [Highest-impact change]
2. [Second highest]
3. [Third]

---

VERDICT
[Ready to ship / Needs revision]

Severity scoring

LOW: 1-3 minor flags. Polish recommended.
MEDIUM: 4-8 flags, or any em dash, or any banned vocabulary. Revise before publishing.
HIGH: 9-15 flags, or multiple structural patterns, or tone fails. Significant rewrite needed.
FAILS PUBLISH BAR: 15+ flags, or any failure on the human/coach test, or corporate-jargon-as-headline phrasing. Do not publish.

Companion skills

For rewriting after audit: use framework-copywriting-hfm for marketing copy, framework-ai-output-hfm for everything else.

· · ·

Section Five

The full master guide.

The complete source document underneath the skills. This is the longest section. Bookmark it and come back to it as needed. It covers the philosophy, the red flags, the language rules, the banned words, the copywriting framework, advanced moves, voice DNA, channel-by-channel tone, and the pre-publish checklist.

1. The core philosophy

Writing that sounds human starts with mindset. Everything in this guide flows from a few non-negotiable truths.

Clarity over cleverness

Good copywriting is clear. Cleverness gets in the way. The deeper work, underneath the clarity, is understanding people. All persuasion is empathy with a call to action. You make someone feel understood. You show them a path forward.

The voice priority hierarchy

Clarity over motivation
Embodiment over inspiration
Truth over hype
Structure over noise
Precision over volume

The meta-question

Before any draft, ask: are we solving a content problem or a clarity problem? Most weak content is a clarity problem. Clarity on who we are talking to, what problem they are living with, and why it matters.

2. Red flags

Before you can write like a human, you need to recognize what makes writing sound like a machine.

Content fails the human test if it...

Uses more than 2 exclamation points
Starts with "In today's fast-paced world..."
Contains a bulleted list with no setup or payoff
Reads like ChatGPT with no custom instructions
Uses passive voice for more than one sentence in a row
Sounds motivational without being actionable
Overuses "I" at the start of consecutive sentences
Includes corporate jargon or feels corporate-neutral
Describes anything as "transformational" without specific evidence

The Human First Media test (all must pass)

Sounds like a real human, not a coach
A high-performing founder would respect this voice
Free of spiritual fluff and Instagram cliches
Grounded in body and leadership
No melodrama, hype, or vagueness
Could be spoken aloud without cringing
Feels contemporary, not 2019 buzzword spirituality
Feels intelligent and lived-in

3. Banned words and phrases

Corporate cliches

synergy, leverage, ecosystem, game-changer, disruptive, impactful, utilize (say "use"), optimize (unless technical context)

AI and coach-speak

"Unlock your potential," "manifest," "I'm so passionate about...," "in today's fast-paced world...," "journey," "unleash," "empower"

Minimizing words

just, simply, basically

Hustle culture

hustling, grinding, "no excuses," "rise and grind"

Filler and hedge words

genuinely, honestly, straightforward, may, might, possibly (when overused), perhaps, "it could be argued"

Shame and urgency tactics

"Summer is coming, are you ready?," "Don't miss out," "Limited time offer," "Act now," "Burn off that weekend"

4. The copywriting framework

The sequence matters. Get the order wrong and the content may be smart but it will not land.

The order

Pain. Start with what the reader is already feeling.
Recognition. Create a mirror moment.
Diagnosis. Name the real problem, not just the symptom.
Solution. Bridge the pain to the answer.

5. Advanced writing moves

Every sentence is a promise. Every sentence either builds trust or breaks it. If it does not earn the next one, cut it.
The best copy has a pulse. It speeds up. It slows down. Long sentences build tension. Short ones release it.
Contrast creates clarity. Before vs after. Wrong way vs right way.
The subtext matters as much as the text. What is implied shapes interpretation.
Every piece is answering a hidden question. If you do not know it, the copy will feel aimless.
Anticipate objections early. Handle resistance while making the case.
The best copy feels like permission. Pressure creates resistance. Permission creates trust.
Specificity creates credibility. Vague claims sound suspicious.
The setup matters more than the punchline. Build the case so the point feels inevitable.
Good copy has layers. Information, emotion, identity.
Silence is powerful. White space matters.
Good copy creates momentum. Each section pulls the reader forward.
Write to one person, not everyone. Precision creates connection.
The best copy feels inevitable. Not pushy. Not forced.
Every word has weight. Choose words for how they feel, not only what they mean.
Break rules on purpose. Fragments. One-word paragraphs. Repetition.
Strong copy has a point of view. Take a clear, defensible position.
You have to feel it yourself. If it does not move you, it will not move the reader.

6. Voice DNA and structure

Sentence construction

Short, punchy sentences mixed with longer, flowing ones. Heavy use of contractions. Moderate pacing with deliberate pauses between ideas. Range from 8-word sentences to 25-word ones within the same paragraph.

The 5-part structure

Start with presence. A short grounding moment. Observational, not spiritual.
Move into clarity. Name what is true. No dramatics, no hype.
Deliver insight. Direct, specific, no metaphors unless simple.
Create an identity shift. Not motivational. Accurate.
Close with clean direction. Honest. Clear. Never poetic.

Signature moves

Reframes. Flip the way someone sees a problem.
Lists of three. Used sparingly, not as a default.
Callbacks. Reference earlier points for cohesion.
Strategic profanity. Rare. Deliberate. Never gratuitous.
Parenthetical asides. Adds texture (like this).

7. Tone by channel

Strategic, executive, thought leadership. Direct insight, no preamble. Proof-driven. Short paragraphs. Clean visual rhythm. Punchy transitions. One clear idea at a time.

Client email

Warm, direct, action-oriented. Short sentences. Clear next steps. Personal but professional.

Proposals and decks

Structured authority. Proof-driven. Data points. Clean progression.

Instagram

Personal, embodied, slightly warmer. Somatic awareness. Real photography, never stock.

Speaking and podcast

Conversational authority. Story-driven. Lived examples. Patient with complexity.

8. Pre-publish checklist

Content check

What is the one feeling this should create?
What is the specific pain?
Does this start with them or with us?
Is there a mirror moment?
Is the real problem diagnosed?
Is the writing readable?
Is the next step clear?
Does the reader feel understood?

Voice check

Sounds like a real human, not a coach
Free of banned words and phrases
Passes the speak-it-aloud test
Free of corporate jargon
Fewer than 2 exclamation points
No passive voice runs

Structure check

Follows Pain → Recognition → Diagnosis → Solution
Rhythm is readable
Opening creates recognition, not explanation
Ends cleanly
You would stop scrolling to read this