Monaze Money

A ledger that holds accounts of seven types, integer-minor-unit transactions, two-legged transfers, computed balances, categories, monthly budgets, statistics, loans, credit cards, receivables, multi-currency and reconciliation, with idempotent import through a developer script. Payees, rules, splits, recurring series and balance projection are specified to extend it additively. Ten screens ship on web, inside an Electron shell on desktop.

Personal finance apps fail in ways their users describe precisely and repeatedly, and almost none of those failures are cosmetic. The category is full of products that model money as a floating-point number, treat a transfer between your own accounts as spending, and store a balance that can quietly disagree with the transactions underneath it. Building another one without first establishing which complaints are structural would have meant reproducing them.

1. 1

   Read the category instead of surveying it

   The research input was 17,804 posts and 14,882 comments across 25 personal finance subreddits, January 2024 to August 2026. The point was not sentiment but recurrence: which complaints show up against every incumbent, which are specific to one product, and which are structural rather than a matter of preference.

2. 2

   Chose a population rather than a feature list

   The research surfaced two large displaced groups: people who lost Mint, and people who stay on Quicken specifically for forecasting. Monaze targets those two. It does not compete with YNAB and Actual Budget on zero-based envelope mechanics, which is a different product for a different person.

3. 3

   Decomposed the product before specifying any of it

   The product was split into sub-projects, each to get its own spec and implementation cycle; three are specified so far: foundation, ledger, and tracking with forecasting. A single specification for the whole product would have been too large to hold, and too large to change.

4. 4

   Proved the integrations before building on them

   The foundation sub-project was scoped to retire three risks: identity tokens authorising database row-level security, a monorepo whose shared packages must resolve under two different bundlers, and a desktop shell holding a session. It shipped one authenticated screen on web and desktop. The mobile app and the production sign-in provider were deferred, so the bundler risk and the identity-provider half of the first risk are still open. Discovering an integration problem after writing features on top of it is the expensive version of the same week.

5. 5

   Borrowed a vocabulary the users already owned

   The shipped ledger was correct and unreadable, and the reason was not styling: it had no vocabulary. Rather than invent one, I asked what this audience already reads fluently. They are people who have kept books (invoices, statements, a ledger that balances), so the interface borrows that language instead of teaching a new one. A green rail you work from, a white sheet you work on, and every screen ending in one action that either balances or does not. Borrowing a mental model the reader already has is the cheapest thing a design can do; building a new one is the most expensive.

6. 6

   Tested the hierarchy in grey before spending any colour

   The redesign was drawn in greyscale first, at 1440x900 and 390x844. It is the only honest way to check a finance interface, because money is the one domain where colour is load-bearing, and if a screen only works when the negative number is red, then nothing on it has a hierarchy. Colour went back on last and stayed rationed: green promoted from an amount colour to the brand, red spent only on amounts, and every state marked by a word, a padlock or a chip as well as a hue.

7. 7

   Redesigned the ledger as bookkeeping software

   The shipped web app is deliberately black and white, with green and red spent only on amounts, and the result gives a nav item, a heading and a €7,850 debt the same weight. The redesign is a design file, not yet in the build: it keeps the model, the figures and the product’s own words, and changes only what outranks what. The two phone screens go further than that. Monaze has no mobile app, so they are a proposal for the native interface the platform decision describes.