Skip to content

Why

This was built for one real situation, by one of the two people in it. That fact shaped it more than any product decision did.

What It Replaced

A folder of markdown files, a long chat transcript, a spreadsheet nobody trusted, and a running argument about which version of a number was current. None of that is unusual. It is roughly what every separating couple assembles in the first month, and it fails in the same way every time: the state lives in two heads and four documents, and the two heads remember it differently.

The insight worth having is that most of the work is not negotiation. It is gathering, reconciling and explaining. Those are the parts that bill at the same hourly rate as the advice, and they are the parts a piece of software is genuinely good at.

The Line It Will Not Cross

There is a version of this app that suggests a fair split. It would be the easiest feature to add and it is the one thing deliberately left out.

The app is allowed to check, to structure, and to notice. It can find the same income entered three different ways. It can turn a sentence into a structured request you confirm before it sends. It can point out that a schedule has drifted from the plan and ask whether the plan is still right. All of that is useful and none of it takes a side.

It does not name a winner on a contested item, summarise one person’s position for the other, analyse tone, or predict what a judge would do. That last one sounds like the most valuable feature and is the most corrosive, because it converts a negotiation into a threat point. A judge would give me more than this is the sentence that ends collaborative separations, and a tool that produces it has taken a side while claiming to be neutral.

Two Users, Permanently

It is not a product. There is no sign-up, and there is no path to one: the two people are values in a database type, which is a design choice rather than an oversight. Building for exactly two known users removes an enormous amount of software, and what is left can be specific in ways a general tool never can.

Whether that generalises is an open question. The arithmetic is jurisdiction-specific, the question bank was reconstructed from one couple’s situation, and the parts that made it good are precisely the parts that assume there are only ever two of you.

A Note on What is Shown Here

Nothing on these pages comes from the live application. Any example data used in screenshots or prototypes is generated from a fictional matter, and a test fails the build if a trace of the real one reaches it. The working app is auth-walled, allowlisted to two addresses, and excluded from search engines.