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 💡

The Voice Was Already There

A few days ago I was rebuilding the Module 8 Framework Guide for the Coaching.com programme. The session teaches a content repurposing workflow: one video in, six pieces out (blog post, email, LinkedIn, short clips, podcast notes, carousel).

A draft sentence in the Guide presented "consistent voice across formats" as something the AI would do. I read it three times and stopped.

Then I rewrote it.

The voice isn't consistent because the AI is clever. The voice is consistent because it starts in the human. The coach records in their voice. That recording is the kernel. Every sub-piece is derived from that kernel. The AI does the distribution. It never touches the voice first.

I changed the sentence to: "You record a piece of content in your voice. That kernel becomes every sub-piece of content." Different sentence. Different teaching. Same words doing different work.

Here's what shifted for me. Most coaches I talk with carry a quiet worry that AI will flatten them. Make their writing sound like everyone else's. The worry has the model backwards. The AI cannot flatten what it never touches first. The flattening only happens when the human disappears from the source. When you skip the recording. When you ask the model to invent the voice instead of carry it.

Source determines output. That is the rule.

Setup matters here. I was using Claude with the Lighthouse mega-prompt, working from a transcript a coach had recorded in their own voice. The output sounded like them. Not because Claude is good (it is, but that's not the reason). Because the input was them. The same prompt run against a generic outline produces generic copy. Same tool, different source, completely different output.

The principle generalises. Anywhere you worry a tool will replace something essential of you, ask whether the tool ever actually touches that thing. Most of the time it doesn't. It distributes whatever you put at the start. If what you put at the start is you, what comes out is you. If what you put at the start is nothing, what comes out is nothing dressed up.

The fear of being flattened is a fear of having nothing at the source. The fix is not less AI. The fix is more presence at the beginning.

“ Voice doesn't arrive with the tool. It arrives with the person using it. — Colin Scotland

AI Prompt That Works ✨

The Five-Minute Voice Anchor

Use this when you want AI to produce content that sounds like you, not like everyone else’s AI. Five minutes of recording, then one prompt. Tangible asset by the end of the session.

Step 1. Open Voice Memos (or any recorder). Record yourself answering one question for five minutes, no script: “What is the one thing my ideal client believes that’s keeping them stuck?” Talk like you’d talk to a friend over coffee.

Step 2. Transcribe. Drop the audio into Claude or ChatGPT and ask for a clean transcript, or use Otter or Whisper if you already have them.

Step 3. Run this prompt:

The Prompt:

Run it once on a question your clients actually ask you. Read the LinkedIn post out loud. If it doesn't sound like you, the transcript was the issue, not the prompt. Remember, You IN = You OUT.

Grounding Practice ☯

Listening for Source

Sit still. Bring to mind one thing you've made recently. A piece of writing, a conversation, a meal, a decision. Something that came out of you.

Now ask: what was the source of that thing? Not the tool you used. Not the format it took. The source. The state you were in when it began.

Were you present when you started? Were you somewhere else? Were you trying to sound like someone you'd read recently, or speaking from where you actually live?

Notice without judgement. The output is always downstream of the source. If the source was clear, the output carried it. If the source was scattered, the output carried that too.

Stay with the question for one minute. Then come back.

Personal Note ❤️‍🔥

When I was nineteen, I had a teacher who heard me before I'd learned how to speak well, queue the bumbling idiot from Liverpool who trips over his words and hates the sound of his own voice. I'd say something half-formed, and she'd say back what I had been trying to say, in clearer words. Not her words. Mine, found.

I lost touch with her years ago. But I think about her every time I write something I trust.

She wasn't translating me. She was making space for what was already there to come through. The voice was mine. She removed the things in the way of it. The insecurities, the uncertainty. She saw me.

That is the work I am trying to teach machines to do this year. And the work I am trying to do for the coaches in the AI Mentorship room next week. And the work I keep failing at when I'm tired and reach for the AI to invent something I have not yet sat with.

The tools cannot make me. They can only carry what I bring.

What I am sitting with this week is how often I forget that. How often I turn to a tool before I have turned to myself. When my AI output sounds generic, it is usually because I was generic when I started.

I do not have a tidy close for this. Just the question I am asking myself: where am I bringing presence to the source today, and where am I outsourcing it?

🎬 Go Deeper

Latest from the blog: Why Obsidian Is the Right Second Brain for the AI Era​

Five jobs a second brain has to do. Most tools fail at three of them. The deeper case for keeping your thinking in your hands.

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 for you. 🙏

Colin

P.S. The next AI Agents Build Sprint runs Friday, May 22nd at 9 am ET. Details at: humanebusiness.ai/build/​