Five things AgeCare will never do
This is a category with a lot of overclaiming: apps that imply clinical judgement, dashboards that imply supervision, and marketing that implies someone is watching when nobody is. Here is what we've ruled out, written down so it can be held against us.
1. Diagnose anything
AgeCare compares a person to themselves. Every chart's shaded band is that individual's own recent range, not a clinical threshold, and the vocabulary is chosen accordingly: slightly above usual, never high. There is no state in the product that means "abnormal".
The escalation signals carry a level, the reasoning behind them, the contributing observations, a confidence rating and a suggested next step, which is usually "mention it at Thursday's appointment". There is no probability and no prediction. If we ever ship anything genuinely predictive, that work starts with a regulatory assessment, not ends with one.
We also state plainly that the product does not diagnose, screen for or measure dementia. Several products in this space are vague about that. Vagueness there is a choice.
2. Monitor anyone covertly
The person being cared for is the data subject, even when a relative is the one paying. That has consequences we accept:
- Location and activity signals require affirmative opt-in and are visible to the person while they're on.
- The senior owns the record. Coordinator status is granted by them, and it's transferable.
- Consent is granular and revocable, and each scope is explained in plain language rather than named after a technical capability.
A monitoring product that the monitored person cannot inspect is surveillance, regardless of the affection behind it. We'd rather lose a feature request than cross that line.
3. Contact emergency services
AgeCare does not call an ambulance, and it must never appear to. What it does is prepare the information a responder needs (allergies, conditions, current medications, contacts, insurance) so it's available in the moment nobody wants to be searching for it, and walk a family through the workflow in advance so it isn't improvised.
If a real emergency feature ever ships, contacting services stays an explicit action taken by a person. Automated dispatch on inferred signals is out of scope permanently: the cost of a false positive lands on emergency services and on the older adult's autonomy, and neither of them agreed to it.
4. Make money from health data
No ads, no data sales, no "anonymised insights" business line. The revenue model is subscriptions from families and contracts with agencies, and that's the whole list. If analytics are ever added, they'll be event-level and won't carry health content, and notification providers will receive a reference rather than clinical text.
Your record stays yours. Export is free, and account closure produces a real export and a real deletion. Where the family's wish to keep history conflicts with the older adult's wish to remove it, it resolves in favor of the older adult.
5. Show a concept as though it were a capability
This is the one that shapes day-to-day decisions most. Where a screen demonstrates something the connected product will do but this build does not yet do, the screen says so in the interface, not in a footnote on a website. Document storage is described as designed for encryption rather than implied to be encrypted. A generated summary says where it came from. A simulated flow says it's a simulation, before, during and after.
It costs us some polish. It's worth it, because the families deciding whether to trust a product with a parent's health can tell the difference between confidence and precision, and because the institutional buyers can definitely tell.
Standing principles
- Never claim a capability that isn't implemented.
- The monitored person is the data subject.
- Deny by default.
- No diagnosis, ever.
- Emergency actions stay human.
- Health data is not a revenue line.
If you find us breaking one of these, write to privacy@agecare.app and say so.