ADHDWhy AI Will Not Replace Your ADHD Coach
I read the newest review of AI-generated ADHD plans as the best possible news for anyone with ADHD who can't afford $60 a week for coaching. Most people read it as a knock against AI. The December 2025 review in BMC Psychology put AI-generated ADHD plans in front of 27 clinical experts and asked them to grade the work. The plans knew their theory. They were built on solid executive-function science, the real stuff coaches train on for years. But the experts kept flagging the same two problems: too generic, and weak on whether a real person could actually pull the plan off in a real week. The reviewers' conclusion was blunt. AI should complement human judgment, not replace it.
It means the cheap, tireless part of the job, the part that used to eat up half a coaching session, can go to a machine. The part that actually needs a human, judgment about your specific life, stays with a human. That split is the correct division of labor, and it saves you money.
What the study actually found
The split is simple. The AI plans got the theory right. Break tasks down, reduce cognitive load, build in structure, all textbook executive-function support. Where they fell apart was translation. A plan that says "block 90 minutes for deep work" is theoretically sound and useless to someone whose ADHD brain has never once sustained 90 minutes of anything. The plan doesn't know your kitchen, your kid's school pickup, your 2pm energy crash, or the fact that you've started this exact task four times this month. A coach knows that, because a coach asked you, or watched you fail at it twice and adjusted.
That gap is the entire difference between advice and support. Advice is generic by definition, it has to work for the average case. Support is specific, it has to work for you, this week, with your actual constraints. AI is very good at the first thing and structurally bad at the second. The reason is plain: it doesn't have your context unless you hand it over, and even then it can't sit with you and read the room.
The plan doesn't know your kitchen, your kid's school pickup, or your 2pm crash. A coach does.
Why this split is good news, not bad news
Think about what coaching actually costs. As of this writing (July 2026), Shimmer's ADHD coaching runs $45 a week for the first four weeks, then $60 a week after that. NeuroSpicy, the biggest ADHD Skool community at around 1,400 members, charges $29 a month with a plan to move to $49 — vendor-page numbers that can change without notice. Real human attention is expensive because it's scarce, and it should go to the part only a human can do.
A coach's real value was never "help me break this project into steps." A generic plan can do a version of that. A coach's value is knowing you avoid the phone because of, say, a bad call in March, or that Tuesdays are dead because that's when your kid has therapy. That's judgment built on repeated contact with one person. No AI plan generated from a prompt has that, and the BMC review is basically confirming what anyone with a good coach already knew.
So instead of asking AI to replace the coach, ask it to do the boring scaffolding the coach shouldn't be spending billable time on anyway. Sorting your brain dump into piles. Shrinking a task to a two-minute version. Holding a restart note so you don't lose your place between sessions. That's the repetitive, low-judgment layer, and it's exactly where AI is strong and cheap. The Yap Method lives entirely in that layer on purpose: YAP to dump the mess, MAP to sort it into five piles, TAP to pick one tiny action, SNAP to save your spot. None of those four steps require a human to be present, and none of them pretend to be a coach.
What the BMC Psychology study found
27 clinical experts reviewing AI-generated ADHD executive-function plans, Dec 2025
What this means for your wallet and your week
If you're weighing whether to spend money on ADHD support right now, this study gives you a useful filter. Anything repetitive, sorting, prioritizing, shrinking a task, holding a note for later, hand that to AI. It's cheap or free and it doesn't need to know your whole life story to sort five items into five piles. Save the money and the human hours for the parts that need someone who knows you: a therapist, a coach, a friend who body doubles with you on a video call. That last one is worth a note because people mistake it for something clinical. Body doubling has almost no formal research behind it, a handful of small papers and theses, no randomized controlled trials. People with ADHD swear by it, and that's a real, honest reason to keep doing it. It's not the same claim as "clinically proven," and nobody should sell it to you that way.
A newer 2026 study of 22 ADHD adults backs up why the human layer matters: ADHD task management is social, not solo. People borrow structure from other people, check-ins, body doubling, someone just being aware of what you're working on. Some participants said reporting progress to an AI already felt like enough of a "someone's there" effect to help them start. That's real and useful, and it's a smaller claim than "AI is your coach now." A CHI 2026 co-design study with 20 ADHD students and 5 experts named three things people need from AI support: help noticing what task you're on, help getting through it once it's started, and help with the emotional noise around both. None of those three require the AI to know your life history.
Use AI for the scaffolding, not the judgment calls
ADHD has real costs. Nobody needs a pep talk to know that. But with the right AI support, some ADHD traits, the fast talking, the pattern jumping, the need to think out loud, can become real strengths instead of things you're constantly apologizing for. The BMC review works as a map. It shows where AI is strong, the repetitive scaffolding, and where it isn't, the judgment calls that need someone who actually knows you. Use it for what it's good at. Don't ask it to be your coach, because it isn't built to be, and now there's a 27-expert review saying so in writing.
The Yap Method is built for the scaffolding layer on purpose. YAP dumps the mess out loud. MAP sorts it into five piles. TAP picks one tiny action, two to ten minutes. SNAP saves a restart note so you don't lose your place. Two override words, SMALLER and SHORTER, if any step still feels too big. It handles the repetitive part so the humans in your life, your coach or your therapist, can spend their time on the part that actually needs them.
Take the whole system with you
The free 43-page ebook: the brain science in plain words, the four steps, the master prompt, and your first seven days.
Prefer it live? Doug teaches this in libraries and meetups — book a 90-minute class →