Rules & automation

Most transactions repeat: the same coffee shop, the same subscription, the same landlord. A Rule tells finzu “when a transaction looks like this, apply this category (and optionally these tags)” — so you categorize a merchant once and every future transaction from it is handled automatically.

A rule matches on what’s actually in the transaction — usually its description — so STARBUCKS #4021 and STARBUCKS #1187 can both match a single “Starbucks → Dining Out” rule, without you having to categorize each one by hand.

Where rules come from

You can write a rule directly, but finzu also watches for the pattern itself: when it notices you’ve manually categorized several similar transactions the same way, it suggests turning that pattern into a rule. You stay in control of what becomes a rule — finzu proposes, you confirm — but it does the noticing for you.

Why rules, not a machine-learning categorizer

A rule is something you can read: “this text pattern → this category.” That’s the whole point. If a category ever looks wrong, you can find and fix the exact rule responsible, instead of wondering why an opaque model made the call it did. Automation here means less repetitive work, not less visibility into how the work gets done.

Next: transactions that move money between your own accounts.