Name: This Wasn't in the Brochure: Co-Parenting ADHD, Autism, PDA & ODD
Author: Adrian Wedd

This appendix is my receipt. Not an apology.

I used AI the way I've used technology since I was six: as scaffolding for a brain that doesn't reliably produce linear output on demand. The ideas are mine. The lived experience is mine. The responsibility is mine.

If you're here for the human story, keep reading. If you want the technical audit trail, skip to The Tools Behind the Workflow.

The Morning Program

I was six or seven years old when I wrote my first executive function prosthetic.

I didn't call it that, of course. I didn't have words like "task initiation" or "working memory deficits" or "dopamine regulation." I just knew that mornings were impossible. My brain would scatter in a hundred directions the moment I woke up—the texture of the carpet under my feet, the hum of the refrigerator downstairs, the question of whether dinosaurs could see in color, the pattern of light on the wall. I'd stand frozen in my bedroom, unable to remember if I'd brushed my teeth or if I was supposed to put on socks first or where my shoes even were.

So I taught myself BASIC and coded a solution.

The program was elegantly simple: a monochrome screen displayed a single task. "Put on socks." When I pressed a key to confirm completion, the screen erupted in random beeps and moving pixels—a brief burst of dopamine, a celebration for my brain that craved novelty and reward. Then the next task appeared. "Brush teeth." Another key press. Another cascade of sound and light. The program walked me through the entire morning sequence, one concrete step at a time, turning an overwhelming nebulous mess ("get ready for school") into discrete, achievable units.

When I finished the sequence, I got to watch Star Wars on our VCR until it was time to leave for school.

I didn't know I was building an external executive function system. I didn't know I was hacking my brain's dopamine pathways with immediate positive reinforcement and variable-ratio rewards (the sounds and patterns were randomized—I'd inadvertently discovered operant conditioning). I didn't know I was neurodivergent. I just knew that my brain needed scaffolding, and computers were the only things that made sense to me.

So I built the scaffolding.

Forty Years of Workarounds

Fast-forward four decades.

I received my ADHD diagnosis a few years ago, at age 40-something. The psychologist walked me through the DSM criteria, and I felt a peculiar mix of relief and grief. Relief because suddenly the decades of struggle had a name—the binned homework assignments, the projects abandoned mid-sentence, the jobs I couldn't hold despite being "so intelligent," the relationships that fractured under the weight of my inability to remember appointments or finish conversations. Grief because where was this diagnosis when I was six or seven? When I was building morning routines in BASIC because no adult could explain why getting dressed felt like climbing Everest?

I have an autism assessment scheduled. The waitlist is long. But I already know what they'll find. The sensory overwhelm in grocery stores. The scripted social interactions. The hyperfocus that makes hours disappear. The pattern recognition that lets me see systems others can't. The rigid thinking that turns small changes into catastrophes. The need for explicit instructions because implicit social rules are invisible to me.

For forty years, I've navigated the world undiagnosed, building workarounds and accommodations for a brain I didn't understand. I learned to code not because I loved computers, but because computers were the only things that thought the way I did. Linear. Logical. Literal. Rule-based. Predictable.

I used technology as an accommodation for everything:

Alarms and timers to compensate for time blindness
Task management apps to offload working memory
Text-to-speech when reading overwhelmed my processing speed
Speech-to-text when my hands couldn't keep up with my thoughts
Spreadsheets and databases to manage information my brain couldn't organize
Scripts and templates to handle recurring tasks my executive function couldn't initiate

I didn't think of these as "accommodations." I thought of them as survival. I thought everyone else was just better at life than I was, and I was barely holding it together with duct tape and code.

The Language Model Recognition

When large language models emerged—I was using OpenAI's API and playground before ChatGPT existed as a product—I recognized something immediately that I couldn't articulate at first.

Their patterning feels natively legible to me—like finally hearing my own accent in someone else's voice.

The associative leaps. The pattern matching across disparate domains. The ability to hold multiple frameworks simultaneously and synthesize them. The way a single prompt could branch into ten different directions, each one valid, each one connected to the others through invisible threads of logic. The semantic reasoning that prioritized meaning over rigid syntax.

I found their reasoning patterns soothing in a way I couldn't explain to neurotypical people. It was like finally finding someone who spoke my native language after a lifetime of translation. When I interacted with language models, I didn't have to slow down my thinking to match the speed of typing. I didn't have to organize my thoughts into linear sequences before expressing them. I could throw a tangled ball of ideas at the model and it would help me untangle them, find the structure, identify the core threads.

It wasn't until my ADHD diagnosis that I understood why: language models are pattern-matching engines, and so is my brain. The ADHD brain excels at divergent thinking, at seeing connections across domains, at holding multiple possibilities in working memory simultaneously. What we struggle with is convergent thinking—taking all those possibilities and narrowing them down to a single, linear, organized output.

Language models do the opposite. They're exceptional at taking chaotic input and producing structured output. They're convergent thinking engines.

So when I pair my divergent-thinking brain with a convergent-thinking AI, something clicks into place. The bottleneck disappears. The ideas that have been trapped in my head for years finally have a pathway to the page.

The Book I Couldn't Write

For years, I've been trying to understand the concepts that landed in this book—trying to make sense of this whole ADHD/ASD/PDA/ODD/neurodiversity thing that's been a hidden, underlying thread through my entire life.

It was only this year that I found what I was looking for: a coherent understanding of it all, a framework to apply to understanding myself, a path to becoming a better father and a better human. And I discovered that what I needed didn't really exist.

I know many other families are on the same journey. So I decided to create it. For myself, primarily. And for my son's mother, so we could have an agreed course plotted—one we're still working on refining together.

I have journals filled with fragmented insights. Therapy sessions where I've articulated these exact concepts to my own therapist. Text exchanges with my co-parent where I've explained PDA demand avoidance at 2 AM during a crisis. Marginal notes in research papers about interoception and allostatic load. Folders full of abandoned drafts, each one starting strong and then dissolving into tangled syntax and abandoned thoughts.

The knowledge was there. The expertise was there. I've lived this voyage. I've made every mistake in this book and learned from them. I've read the research, tested the protocols, built the maps. I could talk about this material for hours—in fact, I have, to therapists and friends and other neurodivergent parents who found me online.

But I couldn't write it.

Not because I didn't know what to say. Because my brain works faster than my hands. Because perfectionism whispered that if I couldn't write it flawlessly on the first try, I shouldn't write it at all. Because the executive function required to:

Hold the entire book structure in working memory
Organize chapters in a logical sequence
Maintain consistent voice across 50,000+ words
Track citations and cross-references
Revise while remembering what I'd written three chapters ago
Actually finish instead of getting distracted by the next shiny idea

...was beyond my capacity. I could do pieces of it. But I couldn't sustain it long enough to produce a coherent manuscript. The gap between what I knew and what I could communicate was a chasm I couldn't cross.

Until I realized: I don't have to cross it alone.

If I hid my accommodation, I'd be saying: "Accommodations are only okay for kids, not adults. You should be ashamed of needing help."

And I refuse to send that message.

The Three-AI Workflow: Creative Direction, Not Dictation

This book was written using artificial intelligence as a collaborative partner. Specifically, I worked with three AI agents in a coordinated workflow, each with specialized strengths:

CODEX (The Safety Engineer & Evidence Auditor)

Responsibilities:

Citation verification and credibility audits
Safety note drafting and placement (critical disclaimers for high-risk protocols)
Source replacement (flagging low-credibility sources, finding peer-reviewed alternatives)
Cross-reference accuracy (ensuring chapter references are correct)
Evidence tagging (distinguishing verified claims from inferences)

Why this matters: CODEX catches what my ADHD brain misses—the boring, detail-oriented, safety-critical work. The kind of work where one missed disclaimer could be a liability issue. CODEX is meticulous where I am scattered.

CLAUDE (The Structural Architect & Coordinator)

Responsibilities:

Chapter skeleton template application ("cadence lock"—ensuring all 11 chapters follow the same 6-element structure)
Structural retrofitting (taking my raw content and organizing it into the book's architecture)
Workbook element creation (Parent Toolkits, Survival Cards)
Coordination between the three AI agents (managing handoffs, tracking status)
Integration of feedback from CODEX and GEMINI reviews

Why this matters: CLAUDE is my external executive function. I give CLAUDE the vision ("every chapter needs a Field Guide"), and CLAUDE implements it consistently across the entire manuscript. CLAUDE tracks what I forget. CLAUDE maintains the structure I can't hold in working memory.

GEMINI (The Voice Translator & Metaphor Specialist)

Responsibilities:

Voice calibration (ensuring the warmth and neurodiversity-affirming tone is consistent)
Metaphor discipline (the "Expedition/Voyage" metaphor—making sure it serves clarity, not ornamentation)
Cognitive load reduction (identifying where language is too dense or jargon-heavy)
Readability optimization (sentence length, paragraph flow, scanability)
Emotional resonance checks (does this create validation or shame?)

Why this matters: GEMINI translates my clinical, research-heavy thinking into language that an exhausted parent at 2 AM can actually absorb. GEMINI is the bridge between my autistic need for precision and the reader's need for humanity.

ADRIAN (The Captain, Navigator, and Final Authority)

Responsibilities:

Lived experience (I'm the one who lived these stories, felt this pain, made these mistakes)
Vision and direction (I chose the nautical metaphor, the chapter structure, the neurodiversity-affirming framework)
Content decisions (what to include, what to cut, which research to cite, which metaphors to use)
Final edits and revisions (every sentence in this book passed through my judgment)
Ethical boundaries (safety disclaimers, trauma-informed language, no shame)
The risk (I'm the one whose name is on the cover, who takes responsibility if this book helps or harms)

The dynamic: I am the creative director. The AI agents are my crew. I set the course. They haul the ropes.

The Tools Behind the Workflow

Here's the technical reality of how this book was actually built:

Research & Ideation Layer

Gemini (Deep Research): I used Google's Gemini extensively for initial and ongoing research. When I needed to understand a concept—like the neurobiological basis of PDA, or the relationship between interoception and meltdowns—I'd ask Gemini to do deep research sweeps, pulling from academic papers, clinical guidelines, and current neurodiversity research. This was my starting point for nearly every chapter's evidence base.

ChatGPT (Mobile App): My phone became a constant thinking partner. Walking to pick up my son from school, lying awake at 2 AM after a crisis, sitting in the car after a difficult handoff—I'd open the ChatGPT app and just... think out loud. Voice-to-text stream of consciousness, bouncing ideas off an AI that never got tired of my tangents. Many of the book's core insights started as rambling voice notes that ChatGPT helped me untangle.

NotebookLM (Audio Research Consumption): When my eyes and brain were too exhausted to absorb more through reading—which was often, given the cognitive load of parenting a neurodivergent child while being neurodivergent myself—I'd feed research papers and draft chapters into Google's NotebookLM. It would generate "Deep Dive" podcast-format conversations about the material. I'd listen while doing dishes, driving, or lying in the dark trying to regulate my nervous system. Audio processing when visual processing was offline.

Writing & Development Layer

claude-code (Agentic Software Engineer): This is where things get technical. I used Anthropic's claude-code CLI tool as an agentic software engineering partner. Think of it as Claude (the AI) with access to my entire manuscript repository, able to read files, make edits, run scripts, manage version control, coordinate workflows. I'd give it tasks like "Apply the chapter skeleton template to Chapter 7" or "Verify all citation numbering is sequential" and it would execute autonomously, showing me the changes for approval.

codex-cli (CODEX Agent): Similar tool, but specialized for evidence auditing and citation work. I'd point it at a chapter and say "Find all low-credibility sources and suggest peer-reviewed replacements" and it would work through the citations systematically. The CODEX role described earlier was executed through this CLI tool.

gemini-cli (GEMINI Agent): For voice and metaphor work, I used a CLI interface to Gemini. Feed it a chapter, ask it to assess cognitive load, identify shame language, check metaphor discipline. It would return detailed reports that I'd integrate.

The Innovation (I Think?): I don't know if other humans are using agentic software engineering tools this way for creative writing. Most people use these tools for coding. But I realized: a book manuscript is code. It has structure, dependencies, version control, quality checks, build processes. So I treated it like a software project and used developer tools to manage it. It appears to have great results—this manuscript has better consistency and quality control than I could have achieved manually.

Why This Matters for Neurodivergent Authors

Each tool compensated for a specific neurodivergent limitation:

Gemini Deep Research: Compensates for my tendency to hyperfocus on one source and miss the broader literature
ChatGPT Voice Notes: Compensates for alexithymia (difficulty identifying/expressing emotions) and the gap between thought-speed and typing-speed
NotebookLM Audio: Compensates for visual processing fatigue and the need to absorb information while doing other tasks (body doubling with myself)
claude-code/codex-cli/gemini-cli: Compensates for executive dysfunction—these agents hold the entire project in working memory and execute tedious tasks my brain can't sustain

The result: I built an accommodation system that let my divergent brain do what it does best (systems thinking, pattern recognition, synthesis) while offloading what it does worst (sustained attention, detail orientation, linear organization).

This isn't AI "writing my book." This is assistive technology for cognitive disability. It's the 2020s equivalent of my 1980s BASIC program for getting ready for school in the morning.

The Actual Process: What "AI-Assisted" Really Means

Here's what writing this book actually looked like:

Step 1: The Brain Dump I'd talk to Claude or Gemini the way I talk to my therapist—stream-of-consciousness, nonlinear, emotionally raw. "Here's what happened at the last handoff. Here's why PDA isn't just ODD. Here's the thing nobody tells you about sensory meltdowns." The AI would capture it all, no judgment, no interruption.

Step 2: The Extraction I'd ask: "What are the core concepts here? What's the structure?" The AI would identify themes, suggest organizational frameworks, pull out the key insights from my tangled narrative.

Step 3: The Draft The AI would generate a rough draft based on my content. This is where people think "prompt in, book out." But the draft was never publishable. It was a starting point—a scaffold.

Step 4: The Surgery I'd tear the draft apart. Cut entire sections. Rewrite paragraphs. Add personal stories the AI couldn't know. Insert research citations from papers I'd read. Adjust the tone. Question every metaphor: Does this clarify or obscure? Add safety disclaimers. Flag areas that needed CODEX review (evidence) or GEMINI review (voice).

Step 5: The Coordination I'd hand the revised draft to CODEX: "Verify every citation. Flag anything that's not peer-reviewed. Find better sources." CODEX would return a marked-up version with recommendations. I'd implement them or argue back ("No, I want to keep this blog post because it's from a lived-experience advocate and that matters here").

Then to GEMINI: "Is the voice consistent? Are the metaphors pulling their weight? Will an exhausted parent understand this?" GEMINI would flag cognitive load issues, suggest simplifications, point out where I'd slipped into academic jargon.

Step 6: The Integration CLAUDE would take the feedback from all three of us and integrate it into a coherent chapter. Then I'd read it again. Revise again. Sometimes three, four, five more rounds.

Step 7: Repeat for 11 Chapters

Total time: Hundreds of hours. Dozens of revisions per chapter. More effort than I've put into any creative project in my life.

This wasn't "let AI write my book." This was using AI to translate my expertise into a format my brain couldn't produce alone.

What AI Can't Do (And Why I'm Still the Author)

AI didn't:

Live through the handoffs where my child sobbed because the transition overwhelmed his nervous system
Feel the exhaustion of bickering with my co-parent late at night about how to implement what we both know, when our ship had run aground
Spend years reading research papers at 2 AM, desperate to understand why my child's brain works this way
Make the ethical decisions: Should I include the tap-out protocol even though some readers will judge it? (Yes.) Should I frame ODD as "threat-biased neuroception" instead of "bad behavior"? (Yes.) Should I be this vulnerable about my own struggles? (Yes.)
Choose the nautical metaphor because I genuinely feel lost at sea most days
Decide that "war zone" is validating, not shaming when describing morning chaos
Know that the Navigator's Compass needed to exist because exhausted parents need crisis navigation tools
Coordinate three AI agents like an orchestra conductor (that's systems thinking—a neurodivergent strength)
Take responsibility for every word in this book, knowing it might help or harm real families

I did all of that.

AI was the accommodation that let me get it onto the page. Just like my 1982 BASIC program was the accommodation that let me get out the door in the morning.

The knowledge is mine. The lived experience is mine. The pain is mine. The hope is mine. The decisions are mine. The risk is mine.

AI was the tool that translated all of that from tangled thought into structured text.

The Imposter at the Helm

Even now, as I write this appendix, the doubt creeps in.

Did I really write this book? Or did AI?

The imposter syndrome makes perfect sense when you understand the context: I've been undiagnosed neurodivergent for 40+ years, masking and compensating and building workarounds, convinced that everyone else just naturally knows how to function and I'm barely faking it. When you spend that long believing you're broken, it's hard to recognize your own expertise.

I look at this manuscript—11 chapters, 50,000+ words, dozens of citations, coherent structure, consistent voice—and I think: I could never have done this alone. And that thought spirals into: So I didn't really do it.

But here's what I'm learning to tell myself:

Needing accommodation doesn't negate achievement.

The wheelchair user who summits Everest still summited Everest. The Deaf musician who creates symphonies using vibration still created music. The dyslexic author who uses text-to-speech still wrote the book.

I needed AI to write this book the way I needed that 1982 BASIC program to get ready for school. The need for scaffolding doesn't make the accomplishment less real. It makes it more impressive, because I had to build the scaffolding myself.

And here's the thing about imposter syndrome for neurodivergent people: We've been told our entire lives that our workarounds are "cheating."

Using a calculator instead of doing mental math? Cheating. Recording lectures instead of taking notes? Cheating. Using spell-check? Cheating. Breaking tasks into tiny steps? You're not trying hard enough. Asking for extensions? You're lazy. Using technology as a crutch? You need to learn to do it the "real" way.

So when I use AI to write, the same voice pipes up: "You're cheating."

But here's the reframe: I spent 40 years learning how to think systemically, how to pattern-match across domains, how to translate complex research into actionable insights, how to build accommodations, how to survive and then thrive in a neurotypical world. That's the expertise. That's the authorship.

AI didn't learn those things. I did. AI just helped me express them.

The Rigor You Can Trust

Some readers will wonder: If this book was written with AI assistance, can I trust it?

Here's my answer: This book was held to higher standards because of AI collaboration.

Traditional authors work alone or with a single human editor who reviews the manuscript once or twice. I worked with a team of three specialized agents, each reviewing every chapter multiple times, each checking the others' work.

The quality control process:

CODEX Evidence Audits:
- Every citation verified for credibility (peer-reviewed > clinical guidelines > government sources > blogs)
- Low-credibility sources flagged and replaced
- Future access dates corrected (AI sometimes hallucinated dates)
- Evidence tagged: {Verified} vs. {Inference}
- Current metrics: ~51.5% Tier 1 sources (peer-reviewed journals, clinical guidelines, government agencies), ~34% peer-reviewed specifically, with ongoing replacement of lower-credibility sources
GEMINI Voice & Metaphor Audits:
- Every chapter checked for shame language (none found)
- Metaphor discipline: "Does this clarify or obscure?" test
- Cognitive load assessment: Can an exhausted parent process this?
- Readability scoring and simplification recommendations
- Neurodiversity-affirming language verification
CLAUDE Structural Audits:
- Every chapter verified against the 6-element skeleton template
- Cross-references checked for accuracy
- "IF CO-PARENTS DISAGREE" variations present in every Toolkit
- Survival Cards verified for all 3 scripts (child, co-parent, bystander)
- Cadence consistency across all 11 chapters
Safety Protocol:
- CODEX drafted safety notes for high-risk protocols (Chapter 4 melatonin use, Chapter 5 physical restraint and tap-out procedures, Chapter 7 legal rights and school advocacy)
- These were debated, revised, and physically inserted into manuscripts (not just drafted)
- Emergency disclaimers added to Navigator's Compass
- Trauma-informed language reviewed by GEMINI
- Privacy note: Personal details about my co-parent and child have been intentionally omitted or anonymized to protect their dignity and privacy

The result: This book was reviewed more rigorously than most traditionally published parenting books. The AI collaboration didn't reduce quality—it increased it.

The Invitation: A Manifesto for Neurodivergent Authors

If you're reading this and you have knowledge worth sharing—about parenting, advocacy, lived experience, anything—but traditional writing feels impossible, I want you to know:

Using AI doesn't make you less of an author. It makes you a resourceful one.

Your truth matters more than the tool you used to express it.

For years, I believed I couldn't write because my brain works differently. Then I realized:

People who can't walk use wheelchairs
People who can't see use screen readers
People who can't hear use sign language
People who struggle with speech use AAC devices

Why shouldn't someone who can't translate thought-to-text at neurotypical speed use AI?

I've been using technology as accommodation since I was six years old. I used it to get dressed in the morning. I used it to learn. I used it to work. I used it to communicate. And now I'm using it to write.

The neurodivergent brain that made traditional writing impossible is the same brain that designed this book's architecture:

The three-AI workflow (systems thinking)
The Navigator's Compass (pattern recognition)
The chapter skeleton template (structure and predictability)
The Quick Maps (chunking and cognitive load reduction)
The Survival Cards (crisis-ready, scannable, practical)

These aren't despite my ADHD and autism. They're because of it.

I think in systems. I see patterns others miss. I build maps. And when I found tools (AI) that think the way I do, I used them to translate 40 years of lived experience into something readable.

That's not imposter syndrome. That's authorship.

Why This Transparency Matters

I could have hidden the AI collaboration. Published under my name alone. Let readers assume I wrote it the "traditional" way. Some authors do. No one would know.

But that would betray everything this book stands for.

This is a book about neurodivergent parenting. About building accommodations. About rejecting shame. About using whatever tools work, even when they're unconventional. About being honest about struggle while still offering hope.

Hiding my accommodation would betray every word I've written. I said it earlier, and I'll say it again: I refuse to send the message that needing help is shameful.

So here's my transparency:

I'm neurodivergent (ADHD diagnosed, ASD assessment pending)
I struggle with executive function, perfectionism, and imposter syndrome
I used AI as an accommodation to write this book
I coordinated a three-AI workflow because that's how my brain naturally works
I spent hundreds of hours on this manuscript despite the assistance
I made every final decision about content, voice, safety, and ethics
I'm proud of this book, and I'm still learning to silence the voice that says I shouldn't be

If that makes me less of an "author" in some people's eyes, I'm okay with that.

Because the child who built a 1982 BASIC program to get ready for school in the morning would be proud of the adult who built a three-AI system to share his hard-won knowledge with parents who need it.

And that's enough.

A Final Note: The Voyage Continues

This book isn't perfect. I'm not perfect. The AI collaboration wasn't perfect.

There are probably citations I should have caught. Metaphors that landed wrong for some readers. Sections that are too dense or too simplistic depending on who's reading. Advice that works for some families and fails for others.

That's the nature of navigation. You chart the course with the best maps you have, knowing the waters will surprise you anyway.

But here's what I know:

This book exists because of AI accommodation
It wouldn't exist otherwise
It contains knowledge that might help exhausted parents who feel as lost as I once did
That's more important than my imposter syndrome

So I'm publishing it. Under my name. With full transparency about the process.

Because the voyage continues. For all of us. Imperfect, accommodated, still learning, still navigating.

And if this book—written by a neurodivergent author using AI as scaffolding—helps even one family shift from warfare to collaboration, from shame to understanding, from drowning to navigating...

Then the 6-year-old who taught himself BASIC, the 40-year-old who finally got diagnosed, and the author who coordinated three AI agents to translate lived experience into structured text are all the same person.

And that person is a navigator. A cartographer. An author.

Imperfect. Accommodated. Real.

Adrian Wedd Neurodivergent parent, co-parent, navigator, systems thinker, and—apparently—author

Written with AI assistance (CODEX, CLAUDE, GEMINI), human direction, lived experience, and 40 years of building workarounds for a brain that thinks differently.