Context

Hormone-related health concerns have shaped both my physical and mental health for a long time. I tried most of the women’s health trackers on the market, looking for patterns I could act on. Every one of them asked me to trade sensitive data for insight I never quite got back.

Problem

Commercial trackers fail in two directions. Apps like Flo bury the useful parts under community feeds and content I didn’t ask for. Apple Health goes the other way: clinical, thin and silent on anything past the calendar and basic symptom tracking. Neither connects a cycle to the things that actually move it, like nutrition, movement and daily life. This makes these apps less optimal for tracking when dealing with conditions like PCOS or PMDD.

Constraints

One non-negotiable: data stays local. That single rule shaped every decision after it.

Process

I started with two assets I already had: a Notion database of nutrition, movement, and symptom guidance built over years, and an export of my Apple cycle history. This was enough for the tool to make grounded predictions instead of generic 28-day math.

I wrote a detailed PRD with Claude, then built with Claude Code, functionality first. Scaffolding before surface. Once the logic held up, I turned to the interface. I wanted ethereal and calm instead of clinical. I collected references on Pinterest, fed the moodboard back into Claude and worked it into a UI prompt.

Outcome

A private companion instead of a data aggregator. It predicts my cycle, tracks symptoms and returns nutrition and movement advice tuned to where I am in it — without a single request leaving my phone.

Reflection

The interesting constraint wasn’t technical. Deciding the tool should feel like a companion rather than a dashboard changed what I built more than the privacy requirement did.