Memoire

Built to remember

Memoire started selfishly. I kept losing track of things that mattered — a meeting from weeks back, where my money went, the task I meant to finish. The name comes from "memoir": something that helps you remember. So I built one place to hold the parts of my life I didn’t want to forget.

The two features I actually live in

It grew a lot of capabilities over time, but two became my daily drivers:

Expense tracking. Everything is categorized, summarized by month, and visualized with charts; amounts can be masked on screen for privacy. After a few months, I wasn’t just recording money — I was seeing it. Patterns I’d been guessing at became obvious, and the way I spend changed because of it.

Task management. Tasks with subtasks, labels, priorities, due dates, recurrence, and search — fast enough that I actually use it instead of abandoning it. One trusted list, synced everywhere and glanceable from a home-screen widget, did the obvious thing: I stopped dropping commitments and got measurably more done.

Those two together are the real product. Tracking money made my finances legible; tracking tasks made me productive. The insight is the payoff — the logging is just the price of admission, and Memoire keeps that price low.

The hard part: making expense logging effortless

The fastest way to kill a finance app is to make people type in every transaction. In Kenya, almost every payment already arrives as an SMS — M-Pesa, Airtel Money, Equitel, KCB, Equity — so the data is there; it just lives in your inbox as unstructured text. Memoire reads those messages on-device and turns them into clean, categorized expenses.

That turned out to be a real engineering problem. M-Pesa alone has half a dozen message shapes — send money, the old and new pay-bill formats, buy-goods, till-only payments — each phrasing the recipient differently. I built a parser that extracts the amount, date, reference code, and merchant from each variant, filters out the noise (money received, reversals, airtime), and guesses a spending category from the recipient. Everything happens on the device, behind an explicit permission, and lands in a review screen where I confirm or tweak each transaction (with duplicate detection) before anything is saved. I prototyped the parsing rules in a quick web build first, then carried the proven logic into the app.

The result is the feature that makes the whole product work: expense logging that costs almost nothing, so the insights actually accumulate.

Built properly, for one user first

The flagship is the Android app, with an offline-first design so it’s quick and works without signal — plus a web companion. It picked up bills, fasting, utility tracking (power, water, fuel), and rich-text notes along the way, but expenses and tasks are the heart of it.

Outcome

• Shipped a cross-platform app (Android + web) end-to-end, solo — published on Google Play
• Built on-device reading of M-Pesa and bank alerts that turns them into categorized expenses, making logging effortless and the insights real
• Built a task system I rely on daily, surfaced everywhere via home-screen widgets and offline-first sync
• Designed an offline-first experience that syncs across devices and keeps working without a connection