Use case · computed on your phone
The useful questions about your own spending are exactly the ones you cannot ask an AI without handing it a bank statement. ShareLess answers them the other way around: the statement stays put and only the answer moves.
A fixed set of money questions, not an open-ended chat about your finances. What you are paying for on repeat. What looks wasteful. What changed month over month. What fees you were charged. How your income compares to your spending.
The arithmetic runs inside the app, on your own files. When your AI asks one of these questions, a card comes up on your phone showing exactly what would be shared before anything goes out, and you approve it or deny it.
Most amounts go out rounded or as ranges rather than exact figures. Merchant names are revealed only for recurring charges and pseudonymized otherwise. Account numbers, balances, and raw transaction rows are never released at all. What the AI receives is an answer, which is a different object from your statement.
The engine is compiled into the iOS app and reads your imported files directly. Every figure it returns is computed from your own statement rather than produced by a language model, which is why this is the one use case on the site we are willing to call enforced.
There is no bank connection. ShareLess is not linked to any account aggregator and does not fetch anything on your behalf. You add a statement yourself through a file picker that takes CSV, plain text, PDF, or a screenshot, and you add a fresh one roughly monthly to keep answers current. Older imports do not quietly become older answers: once your latest statement is far enough behind, the engine stops answering that question and tells you how many days out of date you are instead of returning a stale figure.
The engine and the approval card are both on iOS, which is on invite-only TestFlight rather than the App Store. Your AI reaches it through the ShareLess connector.