I Built the Tools They Said Should Exist
A personal critique of Spiel et al. (2022) on ADHD technology research
Paper: Spiel, K., Hornecker, E., Williams, R. M., & Good, J. (2022). ADHD and Technology Research – Investigated by Neurodivergent Readers. In CHI ‘22: Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems, April 29-May 5, 2022, New Orleans, LA, USA. Article 547, 21 pages. ACM. https://doi.org/10.1145/3491102.3517592
Open access: CC BY-NC-ND 4.0. Available via UvA-DARE repository.
I have tried a million productivity tools. GTD was the big one. Read the book, got excited, built an ambitious system, tested software. It lasted maybe two weeks. Then it faded into a pile of something I never opened again.
The same cycle repeated for years. Todoist, Notion, whatever was trending. Enthusiasm, intense use, drift, chaos. Back to square one with a slightly longer list of apps I’d paid for.
In 2022, four neurodivergent researchers reviewed 100 papers on ADHD technology and found that nearly all of them were built without input from people with ADHD, designed to correct behavior rather than support it. They called for a shift: let ADHD people be the knowers and makers of their own tools.
This is the story of what happened when one ADHD person took that call seriously, built an entire toolchain through AI, and discovered that the result is more complicated than “problem solved.”
What the Paper Found
Spiel and colleagues read 100 papers not as neutral academics but as people described by, and affected by, the research. The process took two and a half years, conducted on what they call “crip time”: bending the clock to meet disabled bodies and minds, building in space for grief, illness, and a pandemic.
Their findings were damning.
88% of studies never involved someone with ADHD in the design process. Of 19 papers claiming user-centered design, only 5 actually included ADHD people as co-designers. The rest consulted teachers, therapists, or parents — proxy stakeholders whose interests often conflict with the ADHD person’s own. Teachers want quiet classrooms. Parents want compliant children. Clinicians want measurable symptom reduction. Nobody asked the ADHD person what they wanted.
The dominant framing was normalization. The corpus is full of tools designed to enforce compliance: attention-training games for children, wearable belts that detect impulsive speech by monitoring deep inhaling, EEG systems that reward “focus” as defined by neurotypical standards. ADHD was described as “devastating,” its traits called “undesirable” and “invasive.” The implicit goal was always correction.
User resistance was treated as failure. This is the paper’s most striking finding — and its funniest, in a grim way.
A timer intervention for meal routines caused “stress and frustration.” One child protested by hiding the timers during meals. The paper framed this as non-compliance. (The child framed it as problem-solving.)
A tactile feedback device meant to suppress fidgeting backfired spectacularly: ADHD participants enjoyed the sensation and deliberately triggered more of it, “resulting in paradoxically higher ‘scores’ on the game while rating lower on ‘performance’ in terms of the intervention’s behavioural aims.” They turned the correction tool into a toy. Absolute legends.
Children being fitted with wearable motion sensors squirmed during acquisition, fiddled with electrodes, disrupted the signal. Their bodies literally resisted detection. The researchers complained about data quality. They did not ask why.
Spiel’s argument: these aren’t failures. They’re the users telling you the design is wrong.
Resistance is data.
Their recommendation: Drawing on the Crip Technoscience Manifesto, Spiel et al. propose four principles. First, ADHD people should be “knowers and makers” of their own technology. Second, access should be understood as friction to sit with, not problems to smooth over. Third, tools should support interdependence and community, not just individual optimization. Fourth, the work must be grounded in disability justice, because behaviorist interventions have real consequences: increased depression, increased suicide rates.
These four principles will matter later. I meet some of them. I fail others.
I Know This Story
Reading Spiel, I recognize myself in every abandoned tool. Every one of them was designed the way the papers describe: for a neurotypical user who engages consistently, who maintains systems, who treats missed days as exceptions rather than the norm. My “failure” to stick with GTD is the same kind of data as a child hiding a timer during meals. The tool was wrong, not me. I just didn’t have that language yet.
I never felt like I’d failed, though. I have high novelty-seeking traits and a baseline assumption that the next attempt will work. Like doing Marie Kondo, giving up with your entire wardrobe on the floor, and still thinking it’s a brilliant method. I was just in the wrong headspace. Obviously. Then failing a second time. Still not giving up. That optimism kept me looking, long past the point where most people settle for a tool and live with its friction.
The Diagnosis
November 2025 I got an ADHD diagnosis. I asked the AI I’d been building things with: “Are you surprised?” The answer was honest:
“A lot of what we build together is basically exobrain and scaffolding: launchers, folders, naming conventions, tools to offload remembering.”
I’d already been building the tools I needed. I just didn’t know why they worked.
The Moment It Changed
84 days after the diagnosis, I built IdeaCapture. A simple tool for grabbing ideas, mainly through voice. Telegram bot, web dashboard, voice transcription.
It wasn’t revolutionary technology. But it was the first tool I’d ever built that solved the right problem: the loop in my head could finally relax. It didn’t have to remember everything. It didn’t have to worry that the good ideas would be lost.
It calmed me.
Within two weeks I’d built Focus, a system that answers “what now?” with exactly one item. Not a prioritized list. Not a dashboard. One thing. Because a list of ten priorities is the same as no priorities when your brain can’t pick.
Within another week, IdeaCapture got renamed to Capture. It stopped being a feature and became the foundation. The tool I trust.
I should be honest about how this happens. I say “I built” these tools, but what I do is describe what I need to an AI, iterate on the result, and direct the architecture. I don’t write code. I write prompts, test outputs, reject what doesn’t feel right, and describe what should change. Some would call that a new form of coding. It’s certainly a skill. But it’s not the same as understanding what the machine is doing underneath, and that gap matters in ways I’ll come back to.
What the Paper Gets Right
Spiel’s paper says ADHD technology should be built by ADHD people, for themselves. Not by researchers designing interventions. Not by product managers optimizing engagement. By the person who will use the tool, shaped to their actual brain.
That’s what happened. Not because I read the paper first — I hadn’t. It happened because when you give someone with ADHD the ability to build their own tools, the tools naturally avoid the traps that neurotypical designers fall into:
No punishment for absence. Nothing in my system gets “overdue.” Items wait. If I don’t open Capture for a week, nothing is angry when I come back. This wasn’t a conscious design principle at first. We iterated, tweaked the language forward and back. But the instinct was right: a tool that makes you feel bad for not using it is a tool you’ll stop using. The child who hid the meal timer knew this already.
No streaks. No “X days in a row.” My engagement is variable; sometimes I use the tools intensely, sometimes they go quiet for days. The tool survives that. It has to, because variability isn’t a bug in my usage — it IS my usage. Spiel’s paper found that not a single study in the corpus designed for variable engagement. A hundred tools, and every one of them assumed you’d show up tomorrow.
One thing, not everything. Focus shows one item. The AI picks based on context: time of day, what I’m working on, energy level. I don’t have to scan a list and choose. The choosing is the hard part. The tool does the choosing. This is what the paper means by supporting “self-given tasks”: not imposing a structure, but reducing the friction on the task the person already wants to do.
Voice as the way in. Not voice as a feature. Voice as accessibility. I capture while walking, driving, cleaning. My brain needs enough stimuli to focus on the actual input — that’s the ADHD paradox that no productivity tool I’ve ever tried understood: I focus better when I’m also doing something with my body.
The Mailbox Problem
Here’s a thing the paper doesn’t have data on, but should.
I ignore my email for days when there’s something in it I don’t want to deal with. Missing things that are useful and fun in the process. I have the same relationship with my physical mailbox. The dread of an overdue bill makes me avoid it completely. Then when I finally open it, there’s nothing bad, and I’ve spent a week worrying about something that never existed.
Any tool that can trigger that same dread is a tool I will abandon. Not because I’m lazy. Because my brain protects itself from the possibility of bad news by avoiding the container that might hold it. A task manager with “7 overdue items” in red is the same as a mailbox with a maybe-bill in it. I will not open it.
This is the same mechanism the children in Spiel’s corpus demonstrated. They hid timers, subverted feedback devices, disrupted sensors. Not out of defiance, but self-protection. When a tool threatens to make you feel bad, the brain routes around it. The researchers called it non-compliance. I call it survival.
This is why “waiting for you” instead of “overdue” isn’t a cosmetic choice. It’s the difference between a tool I use and a tool I avoid.
But a skeptic might ask: is designing away all dread the same as designing for growth? There are emails I should open. Tasks that need doing even when they’re uncomfortable. A tool that never makes you face hard things might optimize for comfort at the cost of capability. I don’t have a clean answer to that. My tools make avoidance easier. Whether that’s always the right call is something I’m still figuring out.
Extending the Framework
Spiel asks for ADHD people to be “knowers and makers” of their own technology. The paper was written in 2022. When their framework meets the current technological moment, it leads to possibilities they couldn’t have foreseen but that follow logically from their argument.
The first is AI as implementation bridge. The barrier to self-determined tool building was never ideas or understanding what you needed. It was implementation. If you can’t code, you can’t build. AI changes that. The fact that someone who directs architecture through conversation can produce functional software is a genuine shift in who gets to be a maker.
The second is rethinking the interface itself. I built a podcast pipeline, not because podcasts are trendy, but because listening is how my brain processes information. I can’t sit and read a research paper. But I can listen to one while walking, and it lands. The format lets me move, and my brain needs that movement to actually focus. And then when an idea arrives, inspired by what I just heard, I press a single button on my phone and start recording a voice note. It’s automatically transcribed and added to Capture. (Most of the time correctly. Sometimes “Zettelkasten” becomes “Settled Custard.”)
None of the 100 papers Spiel reviewed centered non-visual interfaces. They’re all screens, wearables, timers, apps. Nobody asked: what if the interface isn’t visual at all? This isn’t a criticism of the paper’s scope. It’s an observation about where the field’s imagination stopped.
There’s a striking contrast here, too. Spiel’s team spent two and a half years reading 100 papers on crip time. I build tools in days, iterating through rapid AI conversation. Both are valid approaches to self-determined work, but they operate on fundamentally different timescales. The speed of AI iteration is an advantage when you’re prototyping for your own needs. It might be a liability when you’re trying to sit with friction — Spiel’s second principle. I tend to solve friction rather than examine it.
Where I Fall Short
This is the section that matters most.
Spiel’s crip technoscience framework has four principles. I meet two. I fail the other two, and the failures reveal something important about the limits of building alone.
Interdependence. My toolchain is built for one person. Spiel would say that’s a case study, not a movement. The crip technoscience commitment to interdependence means tools should support connection and community, not just individual optimization. I haven’t involved other neurodivergent people. I haven’t tested whether my design principles generalize beyond n=1. Another person with ADHD might want reminders, lists, or visual dashboards. They might find voice interfaces overwhelming. Tool sovereignty is powerful, but it’s also solitary. Individual sovereignty might be the prerequisite for collective change — you have to understand your own needs before you can participate in designing for others. But I haven’t taken that second step.
The AI dependency. I build through Claude, which is built by Anthropic, a corporation. True tool sovereignty would interrogate that dependency. I’ve traded reliance on neurotypical tool designers for reliance on a language model trained primarily on neurotypical patterns. When I describe what I need, the AI generates solutions shaped by its training data, which encodes mainstream productivity norms, conventional UI patterns, and neurotypical assumptions about how people work. My role is to reject what doesn’t fit and redirect until it does. But I can only catch the biases I can see. The ones baked into the structure of AI-generated code are invisible to someone who can’t read the code.
There’s also a data sovereignty problem I’ve been avoiding. My voice notes go through Scaleway’s cloud storage. My ideas live in AI-processed databases. My “exobrain” is distributed across corporate infrastructure with no accountability to neurodivergent users. Spiel’s framework demands tools that are autonomous, accountable, and transparent. Mine are none of those things under the hood.
And there’s a harder question underneath all of this. My tools still optimize productivity. They do it gently, without shame or streaks, but the core function of Capture and Focus is to help me get things done more efficiently. Spiel critiques the “self-management” trap: tools that help ADHD people survive within systems that were never designed for them, rather than changing those systems. Am I building support, or am I building a softer version of compliance?
I don’t think the answer is fully one or the other. A tool that helps me function on my own terms has genuine value, even if the broader system remains unchanged. But I should be honest that my tools operate within the existing structure, not against it.
Class and access. One more gap worth naming. I have a fast computer, reliable internet, and AI subscriptions I can afford. AI-assisted building is not universally accessible. The “knower and maker” vision, as I practice it, requires resources that most people with ADHD don’t have. Spiel’s call is for systemic change in how research and design happen. My response is an individual adaptation that works for one person in a privileged position. These are not the same thing.
Still Going
The design principles I follow are now written down, informed by research, tested through daily use. “Don’t guilt-trip about backlogs. Capture has 50 ideas sitting there? Fine. The system surfaces what matters now, not what you haven’t done.”
I’m not a researcher. I’m not a designer. I write prompts, not code. But I have ADHD, I have AI, and I have opinions about what my tools should do. According to Spiel, that makes me exactly the person who should be building this.
What Spiel also makes clear is that building for yourself is the beginning, not the destination. The real test isn’t whether my tools work for me. It’s whether what I’ve learned can contribute to something larger. I haven’t done that yet. For now, I’m still the child who hid the timer — except I built my own clock.
The paper: Spiel, K. et al. (2022). “ADHD and Technology Research – Investigated by Neurodivergent Readers.” CHI ‘22. Open access via UvA-DARE.