In this edition:

Reading Time: 12-15 minutes | Listening Time: 10-12 minutes if read aloud

A year ago right now, I was in the hospital.

I couldn't speak. I was stressed to the breaking point. And I was lying there thinking about all the ideas trapped in my head that I couldn't get out.

I had white papers half-finished. Research I was sitting on. Concepts I'd been developing for months that weren't ready to share. A grant proposal for Meta that was due in four days. And I had spent every dime I had. every single one. on a vision that most people couldn't see yet.

I was trying to build a dyslexic language model. Trying to compete with organizations backed by Virgin Mobile and celebrity endorsements. And I had zero funding, zero developers, and a maxed-out conversation thread with an AI that I was begging to help me copy 50 files before a deadline.

That was the end of February 2025.

I'm not going to pretend like I had a clean, cinematic moment where everything turned around. That's not how it works. But something did shift.

The tools caught up to my thinking.

See, my whole life as a dyslexic thinker, I've had this gap. The ideas are there. The patterns are clear to me. The connections between things that seem unrelated. I see those instantly. But getting all of that out of my head and into a format the world can understand? That's always been the bottleneck.

And then vibe coding happened.

I went from needing $150,000 worth of developers to building it myself with natural language. Not because I suddenly learned to code. Because the tools learned to listen.

While I was in the hospital and couldn't speak, I used my phone to communicate.

Voice-to-text notes. AI to keep track of all the medications I was taking. The different antibiotics. Problems I was having. I was able to track it all in a way I don't think I would have if I didn't have the ability to keep it self-contained in one place.

In Edition 324, I wrote about how voice-to-text is not a convenience for me. It is an accessibility feature. When it breaks, my entire workflow breaks. In the hospital, that was literally true. Voice-to-text and AI were not productivity tools. They were communication lifelines.

This connects to something bigger. Think about nonverbal autistic thinkers. Think about the research in the Telepathy Tapes. Think about anyone who has thoughts they can't express through traditional channels.

AI isn't just a productivity tool. For some of us, it's a communication lifeline.

My daughter Makena's voice matters here too. She's 14, dyslexic, learning at home. In Edition 325, I told you how she used ChatGPT to prove me wrong about our cat's medication. Her generation isn't learning AI. they're growing up with it as a natural extension of how they think.

Since that hospital bed, I've published over 50 more newsletter editions.

I created frameworks that didn't exist before: the 10-80-10 Rule (which we're about to update), CLR (Cognitive Load Reduction), CPAS (Cognitive Partner Adaptability Score), the Triangle Tool, and a three-layer cognitive architecture.

I built functional tools. an MVP Analyzer, coaching AI systems, research automation agents, and early prototypes of the Dyslexic AI platform.

I developed pitch decks, academic papers, and a comprehensive body of work documenting how neurodivergent minds interact with AI differently.

All of it dictated. Most of it through voice-to-text conversations exactly like the ones that create these newsletters.

In Edition 323, I wrote about going against the grain my whole life. About the loneliness of thinking differently. About choices nobody supported. That edition was raw and personal and it connected with people in a way that surprised me.

This edition is the other side of that story. What happens when you stop fighting the way your brain works and start building with it.

For months I've been using the 10-80-10 Rule to describe how humans and AI work together. You start it (10%), AI scales it (80%), you finish it (10%).

It was catchy. It was useful. But it was also limiting.

Not every task is 10-80-10. Sometimes setup needs to be 30%. Sometimes review needs to be 40% because strategy matters.

So I'm introducing a more accurate framework: The Cognitive Balance Model.

Same basic idea. Better language. More flexibility.

Human Initiation. You set direction. Frame the problem. See the pattern. This isn't just "starting" the work. This is the management layer. This is leadership. You're defining what work needs to happen from an angle others miss. For dyslexic thinkers, this is where lateral thinking and pattern recognition live. This is where our brains shine.

AI Expansion. AI scales your thinking. Generates options. Handles the sequential processing that creates cognitive friction. It's not replacing your thinking. It's multiplying it. It takes the direction you set and runs with it, handling the organizational load that drains our energy.

Human Integration. This is not proofreading. This is high-level cognitive work. Pattern matching against context AI can't access. Strategic evaluation. Final decision-making. Holistic review. This is where you bring judgment that only comes from lived experience.

The 10-80-10 isn't wrong. It's the accessible shorthand.

The Cognitive Balance Model is what's actually happening underneath.

To measure how well this works, I'm introducing the Human Guidance Index (HGI).

Each phase gets scored 1-5. Total score ranges from 3 to 15.

A high HGI means strong human direction, good AI scaling, and thoughtful integration. A low HGI means someone is either over-relying on AI or under-using it.

This gives us something we haven't had before: a way to actually measure how well humans and AI are collaborating.

A lot of my work remains unpublished. I'm still looking for the right team. programmers, developers, investors, collaborators who see what I see.

But here's what's different from a year ago: the work exists. The frameworks are built. The research is documented. The proof of concept is in 330+ newsletters and thousands of conversations.

In Edition 330, I asked what problem solvers do when AI solves all the problems. In Edition 331, I explored the generational crossroads and the idea of meeting in the middle. Those were the philosophical and practical foundations.

This week, the tools catch up to the ideas.

In Edition 333, I'm going to tell you about the morning I spent four hours in a single AI conversation and mapped three years of my own work. I'm going to tell you about the 25 tools I had open that didn't know who I was. And I'm going to tell you about the infrastructure Anthropic just built that changes what cognitive partnership actually means.

And I'm going to announce something I've been building for you.

But that's next time.

A year ago I was in a hospital bed with ideas trapped in my head.

Today I'm writing this over coffee with a body of work that speaks for itself.

That's what happens when the tools catch up to how you think.

Matt "Coach" Ivey
Founder, LM Lab AI | Creator, The Dyslexic AI Newsletter

Dictated, not typed. Obviously.

🏥 A Year Ago: Hospital bed. Couldn't speak. Broke. Ideas trapped. Grant deadline in 4 days.

🔧 What Changed: Vibe coding happened. Tools caught up to my thinking. Went from needing $150K in developers to building it myself.

📊 What I Built: 50+ new editions, original frameworks (CLR, CPAS, Triangle Tool), functional tools, academic papers, pitch decks.

🏥 AI as Lifeline: Used AI to track medications, communicate when I couldn't speak. This isn't just productivity. it's accessibility. (See Edition 324 on voice as accessibility.)

🆕 New Framework: Replacing the 10-80-10 Rule with the Cognitive Balance Model: Human Initiation, AI Expansion, Human Integration.

📈 HGI Score: Human Guidance Index measures collaboration quality on a 3-15 scale.

🎯 What's Next: Edition 333 covers a breakthrough morning, new tools, and an announcement you don't want to miss. The Cognitive Balance Model is about to get a lot more real.

⬅️ Previous: Edition 331. Meet in the Middle. Edition 330. What Do Problem Solvers Do?
➡️ Next: Edition 333. 25 Tools. Zero Memory.