The contradiction at the center
ADHD often craves novelty, variety, and unpredictability. Autism often craves routine, predictability, and known stimuli. A person who has both lives with that tension on a clock cycle that is hard to predict from outside.
Most productivity apps are built around one of those poles and lose the other. Habit-stacking apps work for the autistic side and feel unbearable to the ADHD side. Gamified novelty apps work for the ADHD side and feel intolerable to the autistic side. AuDHD adults churn through both.
Three patterns we built for
Pathological Demand Avoidance (PDA). A subset of autistic adults experience high anxiety in response to demands, including demands phrased as suggestions, including demands made by themselves to themselves. The clinical literature on PDA is still developing (Newson et al., 2003; O'Nions et al., 2018), but the lived experience is consistent: imperatives trigger shutdown.
KickMint's adaptive prompt tone, when the user reports PDA-aware framing helps them, switches the breakdown prompt's instruction style from imperatives ("Open the document") to declaratives ("Opening the document is the first action") and softens the Stuck recovery to "Here is one option you could try." Same content, different cognitive load.
Sensory overload. The autistic side often runs out of bandwidth in sensory-noisy environments. The ADHD side often does not notice this happening until it is already too late. The product answer is Stealth mode: a single tap that dims animations, mutes notifications, and switches the UI to a high-contrast minimal palette. It is a self-regulation lever the user reaches for, not an algorithm trying to predict their state.
Novelty rotation. ADHD craves variety; autism craves consistency. The Novelty Nudge feature lets the user say "I am bored of my current task," and quietly rotates Pick One toward a different category from the one the user worked on yesterday. The autistic side keeps the same surface; the ADHD side gets a fresh face.
Why the on-device AI matters specifically for AuDHD
AuDHD adults disclose more in private writing than in shared tools. That is not a soft observation; it is a practical fact about masking and trust. The breakdowns and Stuck recoveries that the on-device AI produces are most useful when they are honest, and they are honest when nothing about the system makes the user perform.
The model lives on your phone. It downloads once, about 900 megabytes, then it never talks to us again. There is no cloud inference for any AI feature. The optional encrypted sync (Pro) is end-to-end encrypted; the server holds ciphertext for at most 90 days and has no key material.
What we did not build
We did not build a diagnostic. KickMint does not screen for ADHD or autism, and never asks "do you have AuDHD?" as a gateway question. It asks about preferences (PDA-aware tone, sensory mode, novelty rotation) that are useful regardless of identity, and the user opts in.
We did not build a one-size-fits-all profile. The adaptive layers (energy, capacity, optional cycle phase, sensory mode, PDA tone) compose. The user can switch any one of them on or off without losing the rest.
If you want to read more
- Antshel, K. M., et al. (2016). The comorbidity of ADHD and autism spectrum disorder. Expert Review of Neurotherapeutics, 16(3), 279-293.
- Hours, C., Recasens, C., & Baleyte, J. M. (2022). ASD and ADHD comorbidity: What are we talking about? Frontiers in Psychiatry, 13.
- O'Nions, E., et al. (2018). Identifying features of pathological demand avoidance using the Diagnostic Interview for Social and Communication Disorders. European Child & Adolescent Psychiatry, 27(4), 407-419.
The full bibliography is at /science. The privacy policy is at /privacy. The medical disclaimer is at /medical-disclaimer.