The brief was simple to state: replace handwritten revenue and expense logs with something that calculates profit and loss automatically. The hard part wasn't the calculation. That's arithmetic. It was building something the client would actually use every day without training.
The temptation on a finance tool is to add structure: categories, tags, multi-currency support, export formats, permission levels. Every one of those is a legitimate feature. Almost none of them were needed for this client on day one, and every one of them would have added a decision the client had to make before they could log an entry.
So the interface was built around a single loop: log an entry, watch the number move. Everything else was cut or deferred until there was a real reason to add it back. Profit and loss is calculated continuously rather than at month end, which changed the tool from a record-keeping chore into something the client could glance at any time.
The result wasn't a technically impressive system. It was a system that eliminated handwritten logging entirely and gave the client their time back every single month. That's usually the actual brief, even when the stated brief is about a calculation.