Health templates

Keeping a record, not being diagnosed

Almost everything useful in personal health software is a record: what you did, when, how much, and how you felt afterwards. The value shows up later, when somebody asks how long this has been going on, or when a pattern that was invisible day to day becomes obvious across eight weeks. Everything in this category is that, and nothing in it is medical software, diagnosis or advice.

The reason people keep starting these records and stopping is almost never motivation. It is that logging costs more than the log is worth on any given day. A tracker that takes twenty seconds and a decision is a tracker you will abandon in a fortnight, however good its charts are.

The recurring shapes

Habits are the most common: a small set of things you want to do most days, and a streak or a count that makes the doing visible. Medication and supplement logs are the highest stakes, particularly in a household where more than one person might have given the dose. Symptom journals matter for anything intermittent, because the thing a clinician actually wants is dates and frequency, and nobody remembers those.

Then there are the ones that straddle health and everything else: a meal plan, a training log for a specific race, a sleep record kept for a few weeks to answer one question, a recovery log after surgery, a shared record of an elderly parent's appointments and who is taking them.

Why the app store version usually stops fitting

Consumer health apps are heavily built and mostly free, which should be a good deal and often is not. Free means the business model is elsewhere, and health data is the least comfortable data to have an elsewhere. They are also built around a single user, which breaks the moment the record is about somebody you care for rather than yourself, or the moment two people share the caring.

The other pattern is a paid app that does one narrow thing beautifully and refuses to do the adjacent thing at all, so a person with three small records to keep ends up with three subscriptions and three notification settings. And the export, when it exists, is generally an image of a chart rather than the data underneath it.

The case for building your own

A record you built is a record shaped like your actual question. If you are tracking one symptom against one variable for six weeks, that is the whole app: two fields and a date, with nothing else to maintain. It is private to you and to anyone you deliberately invite, sitting in your own installed copy rather than in a product with a growth department. And it is easy to abandon, which is a feature, because most of these records should end when the question is answered.

The AI half lowers the logging cost, which is the thing that actually decides whether a record survives. Saying what happened is cheaper than filling a form, and asking how often something has occurred this month is cheaper than counting. Being honest about the limits: nothing here connects to a wearable, a scale or a glucose monitor, nothing here alarms or reminds unless you ask your AI to add a scheduled reminder, and the data is private but not encrypted so that only you can read it.

The nearest ready-made start

The closest first-party template is Baby Care Log, which is a health-shaped record in every respect that matters here: append-only entries, an event type, a duration, a caregiver stamped by the server rather than typed, and a daily summary written by the AI. If you are keeping a medication or symptom log for anyone, it is a better starting point than a blank page, and asking your AI to reshape it around your own event types is a short conversation.

Otherwise the useful move is to describe the specific question you are trying to answer and let your AI build the smallest thing that answers it. See also family templates for the caring side and household templates for the routines around it. None of this is a medical device, and none of it should replace talking to a clinician.

The limits, stated plainly

Health records are exactly the wrong place for a vague privacy claim, so: an installed app is private by default and only you and the people you invite by email can open it. It is not end to end encrypted. The rows live in a database on Homespun's servers and an operator could technically read them. That is the same posture as most hosted software, it is simply usually left unsaid, and for a symptom journal you should decide about it deliberately.

It is also not connected to anything. No wearable, no scale, no glucose monitor, no pharmacy, no clinic and no health record system. Everything in an app here got there because a person or their AI put it there, which keeps the record simple and means it will never fill itself in.

And it will not chase you. There is no reminder, no streak notification and no push, which makes it a poor habit-nagging tool and a decent record. If a scheduled prompt would genuinely help, your AI can add one to the app. If you are keeping a record for a clinician, the practical advice is to keep it simple enough that you actually keep it, and to bring the dates and the frequencies rather than the app.