profile

The Intelligence Briefing

The score said easy


Hello Reader,

Each week, I share one insight, one prompt, and a short grounding practice to help you grow with AI in alignment, not overwhelm.

This is for conscious entrepreneurs who value substance over shortcuts.


AI Insight That Matters πŸ’‘

The Score Said Easy

Last week I built a way to check whether the guides I write for coaches are easy to understand. I built it in Claude Code, and the first check is a standard readability score, the kind that turns your writing into a school grade.

One of my guides came back at a fourth-grade reading level. By the numbers, it was easy to read and understand.

Then I set Claude up as a different kind of reader. Someone reading in their second language, who joined late, who is busy and will read it once. I gave it the guide and nothing else, and asked it to tell me the one thing the guide wanted it to do.

It couldn't.

A readability score counts two things, how long your sentences are and how long your words are. That's all it sees. Short sentences pass, even when each one carries three ideas the reader has to unpack. A label like "repeats codified" is two words and scores beautifully. One of my lines talked about building your marketing "on rock", and the fresh reader pictured a map drawn on a rock. Five short words, a perfect score, and the meaning was gone.

So here is the one thing I took from it. A score can tell you a sentence is short. Only a reader can tell you it was understood.

Indulge me, stop for a second and consider what you've just read. What did you hear?

This goes well past course guides. Your welcome email, your proposal, the page that explains what you do. We check our writing with tools that count things, a word count, a grammar check, a score, and they tell us it's fine. None of them can tell you whether someone who isn't you knows what you're asking of them.

It's the communication gap.

AI pulls in both directions here. It tends to write short, confident sentences full of labels, which score well and can still lose people. It can also play the fresh reader, if you tell it who to be and ask it to say back what it understood.

I get sent text and I know instantly it was written by AI because I have to read it four times and even then I can't understand it.

The test is simple enough to run on anything you write. Hand it to someone and ask them what they hear. If they can't tell you in one sentence, it isn't finished.

β€œ
Grammar check can tell you if a sentence is accurate. Only a reader can tell you if it is understood.
β€” Colin Scotland

AI Prompt That Works ✨

Read it Like a Stranger

Use this on something you're about to send to people who don't live inside your head. A welcome email, an onboarding guide, a proposal, the About page on your website. Run it in Claude or ChatGPT. It reads your document as a newcomer would, tells you where that newcomer gets lost, and then gives you a rewritten version you can send today.

The Prompt:

I'm going to paste a document I wrote for other people. Read it as this person. English is their second language. They're busy. They haven't seen anything that came before. They will read it once.
​
After I paste it, answer three questions as that reader, in plain words.
What is this?
What is the one thing it asks me to do?
Where did I get lost, or have to guess?
​
Be honest. If you can't answer the second question, say so.
​
Then rewrite the document so that reader could answer all three in one read. Keep my voice, my words and the names I use for things. Open with one plain sentence that says what this is and what to do. Put the plain meaning before any image or metaphor. Explain any term the first time it appears. Show me the rewrite, then list what you changed.
​
Wait for me to paste the document.

Run it on the next email you were about to send this week, before you press send, and see what the stranger says it asks for.


Grounding Practice ☯

Hearing it Fresh

Sit somewhere quiet for five minutes. Let your breathing settle on its own.

Bring back one thing you said to someone this week that you thought was clear. An instruction, a request, an explanation.

Now picture their face as they heard it. Stay with what they heard, not what you meant. Notice where they might have paused, or nodded without following.

If you feel the urge to explain it again, let that go. There is nothing to fix right now.

Rest for a moment in their side of the conversation.

Finish with three slow breaths, and carry a little of that listening into your next conversation.


Personal Note ❀️‍πŸ”₯

β€œIn language it is simply required that it convey the meaning.”
- Confucius

On a call last Thursday, my business partner Rosario and I were talking about where our recent traction had come from. Two examples were a new partnership and a podcast that someone shared with their whole community. We agreed nearly all of it started with somebody seeing something we'd posted.

We'd been posting regularly for about two weeks, and then we stopped. Neither of us likes posting, and we both know it works.

That's an old problem for me. In 2023, right after finishing a course, I wrote in my journal, "I like creating, I like editing my creations... I don't like the marketing of them, and me a marketer?!" Once the making was done, the polishing and selling went stale on me.

Confucius says language only has to convey the meaning. That's a smaller job than polishing, which I suspect is what I've been avoiding.

We agreed to pick it back up. I don't know yet if it will stick, because last time it didn't.

What do you already know works, and keep putting off? I'd like to hear what it is.


🎬 Go Deeper

AI drafts leave fingerprints, and readers can sense them without being able to name them. This piece covers the ones to look for and a ten-minute read-aloud pass through one page of your website.

Read: Your AI Leaves Fingerprints​

On YouTube

Step-by-step walkthroughs and tactical implementation for conscious AI integration.


Thanks for being here. If this resonates, share it with someone who would benefit from hearing this.

Reply and let me know what landed. πŸ™

Colin

P.S. The next cohort of the AI Accelerator at Humane Business starts Thursday, October 1. Up to five people from your business, building your own AI systems with Rosario and me alongside you. If October is too soon, the following cohort starts in February. Details at humanebusiness.ai/accelerator​

The Intelligence Briefing

Weekly field notes on scaling intelligence without losing your humanity. Get the strategic frameworks, unfiltered insights, and human-first AI methodologies.

Share this page