Nine screens, and nothing on any of them tells you what to do.
This is the real design, running in the page. Not a screenshot, not a video. Drive it here, or read below what each screen is for and why it is that shape.
A fixed sample shelf. The effects shown are the kind an Australian PI lists
Today
What is due, what is not, and one question waiting if the shelf produced one. No score, no streak, no ring to close.
Steady mode lives here. It keeps the shelf, the reminders and the record, and drops the attributional loop, for anyone whose illness makes "what else could be doing this" the wrong question to be asked every day. You choose it. It is never inferred from your data, because inferring it would be the diagnosis this refuses to make.
Today
Tue 3 FebIs the St John's Wort worth raising, given the sertraline?
The shelf
Prescriptions, supplements, food and the rest, in one list. Getting things onto it: share the eScript token you already received by SMS, connect an Active Script List, or type it.
The QR in that token encodes an identifier, not the medicine; resolving it needs a live call to the Prescription Delivery Service. Anything claiming to read a script offline is guessing, and the app says so on the screen where you share one.
Your shelf
6 itemsLook up what you are getting
You bring the symptom. It matches the text against labelled effects on your own shelf and returns every hit alphabetically. It never ranks by likelihood and it never names a condition. Search by condition is discovery and recommendation, which the TGA treats as advertising a prescription medicine.
There is no unlimited loop and no notification inviting a re-check. Symptom searching is observed to escalate from a benign symptom to a severe rare diagnosis inside one session, and search frequency tracks cyberchondria scores in a dose-response way.
Not sleeping
3 hitsThe question sheet
The only thing this app produces. Your medicine, the labelled effect, the tier, the direction-of-causation reading, and the source with its label version and date, as a short sheet you can hand over or share as a PDF.
It is a question, not a request. It does not suggest stopping anything, does not name a dose, and does not say which of three is most likely. Those are your prescriber's to make, and the app has no basis for any of them.
For your prescriber
I have not been sleeping. Three things on my list document that. Is any of it worth looking at?
Your days
Coverage leads, ahead of any number: how many days are recorded and which ones are missing. The days people skip are not random, so the averages they leave behind are biased optimistic and the app says which direction the bias runs.
When there are not enough days for a pattern, that is what renders: an ordinary row with a reason, not an apology and not an empty screen. When there are, a pattern only shows if it beats about 95% of reshuffles of your own days, which is also how the noise floor gets set.
Your days
19 of 31Help
One tap from every screen, on a pill that is always there. Never triggered by what you typed or how you rated your day. Reading your state and routing you on it is triage, and triage is what pulls the whole app into Class III.
Australian lines with their real numbers and hours. Nothing from your shelf renders here. And it says plainly that police attend about 2 to 3 in every 100 calls, because saying so is what measurably makes people more willing to ring.