Can you measure humanity in content?

_{Day 251/100}

Hey—It's Tim. 

Four months ago, I started asking a question that kept me up at night:

"Can you measure humanity in content?"

Not "does this pass AI detection" (we all know those are broken).
But: "Does this sound like a person wrote it, or a robot regurgitating training data?"

Turns out, you can measure it. But holy hell, is it complicated.

I spent 4 months reading 212 research papers, building 31 prototypes, and nearly giving up 6 times.

The result? VIBE Score.

It launches this week. And no other tool on the planet has anything like it.
Let me show you what we built.

The four signals of human content

Most AI detectors look for "AI patterns." Repetitive phrases. Predictable sentence structures. GPT-isms like "delve" and "utilize."

We did the opposite.

We asked: "What makes human writing sound human?"

After reading 212 research papers on linguistics, cognitive processing, and content reception, we landed on four signals:

I walked you through each letter of VIBE:

Viewpoint — Does your content take a stance, or hedge like a robot?

 Individuality — Does this sound like YOU, or generic "content marketer voice"?

Beat — Does your writing have rhythm, or drone like a terms of service doc?

Evidence — Are you citing "studies show" or "we analyzed 500 blogs and here's what happened"?

If you missed any of those, go back. They matter for what I'm about to tell you.

Because here's the thing:

Understanding V, I, B, and E individually? That's the easy part.

Knowing how they work together? That's where it gets weird.

Why adding these scores together is wrong

Here's what we tried first:

Viewpoint: 85 Individuality: 78 Beat: 92 Evidence: 71

Total: 326 → Divide by 4 → Score: 81.5

Looks good, right? Wrong.

Because that score treats a thought leadership piece the same as a step-by-step guide.

Let me show you why that's broken.

The archetype problem

I ran our VIBE scorer against 500 of the best-performing blogs on the internet.

Not "content marketing approved" blogs. Blogs that drove actual revenue. Demos. Pipeline. And some of the best articles on the internet. Top of Medium. Best of the New Yorker. The minimum standard was 1M views.

Then I grouped them by outcome and structure.

Result: 18 different content archetypes.

Each one has a completely different VIBE pattern.

Here are 3 examples:

Archetype 1: The Manifesto

(Example: "Why top-of-funnel content is for cowards")

Ideal VIBE Pattern:

• Viewpoint: 90+ (needs a STRONG stance)

• Individuality: 85+ (personal, opinionated)

• Beat: 80+ (punchy, varied rhythm)

• Evidence: 50 (some support, but opinion-led)

What this means: If you write a manifesto with low Viewpoint (trying to be balanced), it fails. The whole point is the take.

Archetype 2: The Technical Guide

(Example: "How to set up HubSpot + Salesforce sync in 2 hours")

Ideal VIBE Pattern:

• Viewpoint: 60-70 (neutral, instructional)

• Individuality: 50-65 (some personality, but not distracting)

• Beat: 70-80 (structured, scannable)

• Evidence: 95+ (every step needs proof it works)

What this means: If you write a guide with high Viewpoint (lots of opinions), people get annoyed. They want instructions, not your hot takes.

Archetype 3: The Case Study

(Example: "How Sarah's team shipped 40 blogs in Q2")

Ideal VIBE Pattern:

• Viewpoint: 70-80 (has perspective, but not preachy)

• Individuality: 90+ (story-driven, specific person)

• Beat: 75-85 (narrative flow)

• Evidence: 90+ (needs real numbers, timeline, outcomes)

What this means: Case studies live or die on specifics. "Sarah increased output" = weak. "Sarah went from 4 blogs/month to 40 in 8 weeks" = strong.

How VIBE scoring actually works

When you write content in Penfriend this week, here's what happens:

Step 1: You write or paste your draft.

Step 2: Our system analyzes it and detects the archetype.

• Is this a manifesto? A guide? A case study?

• We don't ask you. We detect it from structure, language patterns, and intent.

Step 3: We score your VIBE against the ideal pattern for that archetype.

Example:

• You wrote a Technical Guide

• Your Viewpoint score: 88 (too high for guides)

• Your Evidence score: 72 (too low for guides)

• Result: 64/100 — because you're writing the wrong VIBE pattern for this type

Step 4: Penny tells you exactly what to fix.

Not: "Be more concise."

But:

Specific. Contextual. Actionable.

Step 5: You edit. Your score climbs. You hit 80+. You ship.

What nobody else has

I've used every content tool on the market.

• Jasper, Copy.ai, Writesonic: Generate content. Don't tell you if it's good.

• Grammarly, Hemingway: Fix grammar and readability. Don't measure humanity.

• Surfer, Clearscope, MarketMuse: Optimize for SEO. Don't care if it sounds robotic.

• Originality.ai, GPTZero: Detect AI. But "passing" doesn't mean it's good.

Penfriend is the only tool that measures:

1. Does this sound human?

2. What archetype is this?

3. Is this the right VIBE pattern for this type of content?

4. What specific edits will push your score higher?

This is built on:

• 212 research papers (linguistics, cognitive processing, content theory)

• 500 high-performing blogs (analyzed for patterns)

• 18 content archetypes (each with unique VIBE fingerprints)

• 8 months of development (31 prototypes, 6 near-quits)

No other tool has this. Because no other tool has tried.

Try this before launch

Here's what I want you to do:

Pick your last blog post (or one you're drafting).

Reply to this email with:

1. The URL or paste the draft

2. What type of content you THINK it is (manifesto, guide, case study, etc.)

I'll run it through VIBE (we're testing the final build this week) and send you back:

• The archetype we detected

• Your current VIBE breakdown (Viewpoint, Individuality, Beat, Evidence)

• The 3 biggest edits that would push you over 80

First 15 replies get this today. After that, I'll batch responses for Wednesday.

Why am I doing this? Two reasons:

1. I want to make sure the detection is accurate. Does our archetype match your intent?

2. I want to see if the feedback lands. Are the recommendations actually helpful, or do they feel generic?

You help me stress-test the final build. You get a free analysis.

Launch details

VIBE Score goes live this week.

For everyone on any Penfriend plan. No extra cost.

Or if you just want your blog analyzed before launch, reply with:

1. URL or draft

2. Content type you think it is

Let's see how close we got.

See you tomorrow.

✌️ Tim "Well yes, but actually no" Hanson
CMO @Penfriend.ai

Same brain, different platforms: X, Threads, LinkedIn.

P.S. What's the one piece of content you're struggling to make sound human?

Reply with the URL (or paste the draft) and I'll run it through VIBE before launch. Tell me what you THINK the problem is.

First 15 replies get a full breakdown today: archetype detection, your VIBE scores, and exactly what to fix.

Penfriend.ai
Write the internet’s favorite articles.