Drop in your monthly statements.
See where your money actually flows.
968 transactions categorized in one pass. Future imports inherit these rules automatically — you only revisit when a new merchant shows up.
Totals, averages, and breakdowns for any window — by category, by source, by month. The same ledger read as a balance sheet instead of a feed, so the questions you actually ask have a number waiting.
Every transaction dropped onto a month grid. The heavy days, the recurring charges, the weekends that cost more than you thought — visible at a glance instead of buried in a list.
Promote any merchant into a rule and every future import inherits it. Set the priority, pick the category — your custom rules always sit above the curated defaults, so a one-time fix stays fixed.
Financial data deserves certainty before probability. Deterministic rules match each transaction against an in-house knowledge base — exact, auditable, repeatable. A proprietary NLP matching layer handles the rest against a curated catalog. Only what remains reaches an external Google LLM, bounded to your category tree. A system that hardens as it learns. A self-reinforcing loop: graph clustering mines recurring transactions, promotes them to deterministic rules, and shrinks whatever reaches the probabilistic model.