Hey friends,

I read something this week that made me put my coffee down. And if you know me, you know I don't put my coffee down for just anything.

Dr. Andy Johnson, a reading specialist at Minnesota State University, published an article called "Dyslexia is NOT a Brain Disorder."

Now before you react to that title, hear me out. Because what he's actually saying is something I've been circling for two years and 327 editions of this newsletter. And I think it changes how we talk about everything we do here.

His core argument? The brain imaging research that gets used to call dyslexia a "neurological disorder" has real problems. Small sample sizes. Limited task types. Over-interpretation of data. And here's the big one: many of the differences researchers see in brain scans largely disappear with the right kinds of instruction and experience.

That means the differences might be about exposure and environment, not permanent wiring.

But here's where I take it a step further. I think the AI revolution gives us a completely new way to frame this conversation.

Why calling dyslexia a "disorder" or a "gift" both miss the mark

A new framing: dyslexia as cognitive resource allocation

Why AI isn't fixing you. It's load balancing your cognitive system.

What Claude Opus 4.6 and 'agent teams' mean for neurodivergent thinkers

Why 2026 is being called the year of multi-agent systems and what that means for us

The vision: AI agents built from the ground up to think like dyslexic minds

I've never been comfortable with the word "disorder." Most of you reading this probably feel the same way.

But I've also never been totally comfortable with the "dyslexia is a superpower" crowd either. Because calling it a superpower romanticizes something that can be genuinely exhausting. And it dismisses the real friction that comes with how our brains process written language.

So if it's not a disorder and it's not a gift, what is it?

Here's the framing I've landed on, and I think it's the most honest and structurally accurate way to describe what's happening:

Dyslexia is a spectrum of cognitive resource allocation.

Your brain has a finite amount of processing power. It has to distribute that power across language decoding, spatial reasoning, pattern recognition, abstraction, memory, and a bunch of other cognitive functions.

In a dyslexic brain, the allocation is distributed differently. Not wrong. Not broken. Distributed differently.

Spelling friction and what researchers call orthographic processing load often correlate with stronger global pattern detection, abstraction, and systems thinking. Not because one magically causes the other. But because neural efficiency is finite. Emphasis in one domain often means tradeoffs in another.

That's not pathology. That's allocation.

Think of it like a budget. You have a fixed amount of cognitive currency. Some brains spend heavily on decoding written symbols, leaving less for big-picture pattern recognition. Other brains, our brains, spend heavily on abstraction and spatial reasoning, leaving less for the symbolic stuff.

Neither budget is wrong. They just prioritize differently.

And here's the part that matters most for what we talk about in this newsletter:

When that allocation is balanced intentionally, it becomes powerful. When it's unsupported, it becomes exhausting.

This is where the reframe gets really interesting. Because once you see dyslexia as a cognitive resource allocation pattern, the role of AI completely shifts.

AI is not remediation. It's load balancing.

AI handles the high-friction symbolic tasks. Spelling precision. Syntax refinement. Linear structuring. Sequential organization. The stuff that eats up massive amounts of cognitive energy for a dyslexic thinker.

You retain the high-leverage capabilities. Pattern detection. Strategic thinking. Abstraction. Narrative synthesis. The stuff your brain allocated extra resources to in the first place.

AI isn't making up for a deficit. It's redistributing workload across systems.

When I use voice-to-text, I'm not compensating for a weakness. I'm offloading the symbolic encoding that costs my brain a disproportionate amount of energy, so I can operate at the speed of my actual thinking.

When a student uses Speechify to hear text read aloud, she's not taking a shortcut. She's reallocating the processing load so her brain can focus on comprehension and meaning instead of burning all its energy on decoding.

When Makena used ChatGPT to structure her argument about cat medication, she wasn't cheating. She already had the knowledge. The AI handled the linear sequencing so she could deploy the pattern recognition and strategic reasoning she's naturally strong in.

Same cognitive budget. Better allocation. That's what AI does for neurodivergent thinkers.

And that framing, right there, is what I've been calling the cognitive partnership model since the beginning of this newsletter. I just didn't have the right language for it until now.

While we're reframing how we understand dyslexia, the AI models themselves are evolving in a direction that makes this load balancing more powerful than ever.

Earlier this month, Anthropic released Claude Opus 4.6. It's the model I use every day. And what they added changes the game for us.

Adaptive thinking.

Previous AI models had two modes: think hard or don't think hard. On or off. You had to tell the model how much effort to put in. Now the model reads the complexity of what you're asking and decides for itself how deeply to reason through the problem.

Think about that through our lens. An AI that adjusts its cognitive effort based on what you actually need. Not a fixed template. Not a one-size-fits-all response. An adaptive load balancer that matches the complexity of each interaction.

Agent teams.

This is the one that really gets me excited. Instead of one AI working through a task step by step, you can now split work across multiple specialized agents that each own their piece and coordinate with each other in parallel.

One agent handles research. Another handles writing. Another handles fact-checking. Another handles formatting. All running at the same time. Talking to each other. Producing a finished result together.

Sound familiar?

It should. Because that's how teams of humans work. A project manager, a developer, a QA person, a designer, each owning their domain and coordinating the whole.

Now imagine those agents are fine-tuned to understand how your cognitive allocation works. Which tasks burn your energy. Which tasks light you up. Where you need load balancing and where you need amplification.

We're not there yet. But the foundation is being laid right now.

If 2025 was the year AI agents became real, 2026 is the year they learned to coordinate.

Everywhere I look in the AI space right now, the conversation has shifted. It's no longer about a single AI that does everything. It's about specialized agents that work together.

Right now, when you use an AI tool, you're having a one-on-one conversation. You ask something. It responds. You refine. It responds again. Helpful, but it's one brain working on one thing at a time.

Multi-agent systems change that. Instead of one conversation, you have a whole team of AI agents, each specialized in a different skill, working on your problem simultaneously.

One agent that's great at research. One that's great at writing. One that's great at visual thinking. One that catches errors. One that simplifies complex language.

Now here's where my brain goes as a dyslexic thinker.

What if one of those agents is specifically designed to understand lateral thinking? To follow the associative leaps that neurodivergent minds make naturally? To not just tolerate nonlinear reasoning but to amplify it?

What if another agent handles the execution side, the part that costs us the most cognitive energy? Taking the big-picture vision and breaking it into sequential steps without losing the creative insight?

What if a third agent translates between cognitive styles? Taking the way I think and structuring it so a linear processor can follow?

That's not science fiction. That's where the technology is heading. Right now.

Some of you know I've been working on a cognitive architecture for dyslexic AI tools. I call it the three-layer system. And it maps directly to the load balancing concept.

The Socratic Layer asks you questions. It draws out your thinking. It doesn't tell you what to do. It helps you discover what you already know. Like the best teacher you ever had. This layer works with your brain's natural pattern recognition instead of against it.

The Strategic Layer takes what the Socratic conversation reveals and builds structure around it. It organizes the chaos. It creates the plan. It handles the sequential execution, the high-friction, high-energy tasks that our cognitive allocation doesn't prioritize.

The Skeptic Layer challenges everything. It pressure-tests the ideas. It asks "what are we missing?" It makes sure the creative leaps actually hold up under scrutiny. This is the quality control agent that turns lateral insight into something defensible.

Three layers. Three cognitive functions. Each one mapped to how neurodivergent minds actually allocate resources, not how the education system says they should.

When I look at what Claude just released with agent teams, and what the multi-agent ecosystem is becoming in 2026, I see the infrastructure for exactly this kind of system.

Not as a feature request. Not as a wish list. As something that can actually be built.

Multiple specialized agents, each handling a different cognitive function, working together to load balance a dyslexic thinker's processing in real time.

Let me connect the dots. Because I think Dr. Johnson's article, the cognitive allocation framing, and the AI convergence all point to the same conclusion.

If dyslexia is a permanent brain disorder, then AI is assistive technology. A crutch. Something you need because you're broken.

If dyslexia is a superpower, then AI is optional. Nice to have. But you should be able to do it all on your own because your brain is magic.

But if dyslexia is a cognitive allocation pattern? Then AI is something entirely different.

AI becomes the load balancer that makes your allocation work at full capacity.

It's not fixing what's broken. It's not supplementing what's already perfect. It's taking a cognitive system that distributes resources in a specific way and adding capacity exactly where the allocation creates friction.

That's not remediation. That's optimization. That's cognitive partnership.

And when I read Dr. Johnson's argument that brain imaging differences largely disappear with the right instruction and experience? That tells me the system is adaptive. Malleable. Responsive to the right tools and the right environment.

AI is the right tool. Multi-agent systems are the right environment. And we're building both right now.

Here's where I'm going with all of this. And I want you to really sit with it.

Imagine a world where you have a team of AI agents that are specifically designed to work with your cognitive allocation. Not adapted from a neurotypical template. Not retrofitted with accessibility features.

Built from the ground up to load balance your specific pattern.

An agent that follows your lateral connections instead of forcing linear structure.

An agent that handles the sequential execution your allocation doesn't prioritize.

An agent that translates your big-picture insights into language that linear processors can follow.

An agent that asks you the right questions to draw out ideas you didn't even know you had.

An agent that runs in the background, researching, organizing, connecting dots, while you do what you do best, which is see patterns nobody else sees.

Running autonomously. As an extension of your cognitive system.

The question isn't whether this will happen. The infrastructure is being built right now. Multi-agent systems are the defining trend of 2026. Adaptive AI models are already matching their effort to each interaction. The building blocks exist.

The question is whether neurodivergent thinkers will be at the table when it's designed.

I intend to be at that table. And I want you there too.

If you're reading this newsletter, you're already ahead of the curve. But here's what I want you to take away from this one.

The language matters. How we describe dyslexia shapes how the world treats us and how we treat ourselves. "Disorder" invites pity. "Superpower" invites dismissal. "Cognitive allocation pattern" invites understanding. Use the language that serves you.

AI is not your crutch. It's your load balancer. When you use voice-to-text, you're not compensating. When you use AI to structure your ideas, you're not cheating. You're redistributing cognitive load so your natural strengths can operate at full power.

The technology is converging. Adaptive AI, multi-agent systems, specialized cognitive tools. All of it is coming together. And the downstream benefits for neurodivergent thinkers are going to be massive.

The biggest barrier isn't the technology. It's understanding. It's teachers who don't know the difference between a student cheating and a student load balancing. It's systems that weren't built for different allocations. It's a world that's still catching up to what neurodivergent people have already figured out.

You are not behind. You are early. If you're a dyslexic thinker already using AI as a cognitive partner, you're not playing catch-up. You're pioneering. The rest of the world is going to realize what you already know: these tools don't replace thinking. They amplify it.

Strengths and frictions emerge from how the system balances language decoding, spatial reasoning, abstraction, memory, and pattern recognition.

When balanced intentionally? It becomes powerful.

When unsupported? It becomes exhausting.

AI is the intentional balance.

And the tools to make it happen are here. Right now. Being built by people who think like us.

If this reframe resonated, share it with someone who needs a better word than "disorder." And if you want to go deeper on the personal stories behind the pattern, check out the last edition where I told you about a conversation over deep dish pizza that I can't stop thinking about.

Matt "Coach" Ivey

Founder, LM Lab AI | Creator, Dyslexic AI

(Dictated, not typed. Obviously.)

🧠 The Reframe: Dyslexia is not a disorder and not a superpower. It's a cognitive resource allocation pattern. Your brain distributes processing power differently. Emphasis in pattern recognition and abstraction often means tradeoffs in symbolic decoding. That's not pathology. That's allocation.

⚖️ AI as Load Balancer: AI handles the high-friction stuff (spelling, syntax, linear structure). You retain the high-leverage stuff (pattern detection, strategic thinking, abstraction). AI isn't fixing you. It's redistributing workload across systems.

🤖 Claude Opus 4.6: Anthropic's newest model introduced adaptive thinking (AI that adjusts its effort to match your needs) and agent teams (multiple specialized AI agents working together in parallel). This is the infrastructure for building custom cognitive tools.

🔄 Multi-Agent Systems: 2026 is the year AI agents learned to coordinate. Instead of one AI doing everything, specialized agents split the work. Research, writing, fact-checking, translating between cognitive styles. All running simultaneously. All load balancing different cognitive functions.

🚀 The Vision: AI agents specifically built for neurodivergent cognitive patterns. Following lateral connections instead of forcing linear structure. Handling sequential execution while you see patterns. Translating your insights for linear thinkers. Running autonomously as an extension of your cognitive system.

💡 The Point: When your cognitive allocation is supported intentionally, it becomes powerful. When unsupported, it becomes exhausting. AI is the intentional support. The tools exist. And if you're already using AI as a cognitive partner, you're not behind. You're early.

Dr. Andy Johnson's article "Dyslexia is NOT a Brain Disorder" is available on LinkedIn and Substack. Whether you agree with all of his conclusions or not, the conversation it opens is one worth having.