Find savings in my spending

Using the Qonto MCP, analyse my outflows over the last 30 days, compare them with the previous 30 days, and tell me exactly what I can cut, downgrade or renegotiate, how much it saves per year, and how to do it this week.

Steps

1. Scope the account. Call get_organization (legal form, country) and list_bank_accounts (accounts, main account). Analyse the main account unless the user names another one. The legal form drives the tax notes in step 7.

2. Pull two windows of debits. With list_transactions (side: debit, per_page: "25", sort_by: "emitted_at:desc", filtered with emitted_at_from and emitted_at_to), paginate until next_page is null for two windows of exactly 30 days: the current one (today minus 29 days to today) and the previous one (today minus 59 to today minus 30 days). Pages of 25 keep each result small enough for any host. Each transaction carries its cashflow_category.name; list_cash_flow_categories gives the full catalog. Keep the total of credits of the current window for context only.

3. Classify every line by nature. Use the Qonto category first, then clean_counterparty_name, then label, then operation_type.

   • Controllable: software and SaaS, telecom, tools, bank and card fees, ads. This is where savings live.
   • Structural: payroll, subcontractors, rent, insurance, taxes and social contributions, mandatory certifications, owner draws. Never "cut", only flag.
   • One-off: travel, large isolated purchases. Excluded from the run-rate. Fees come as separate lines: operation_type: qonto_fee with reference: fx_card is a foreign-exchange fee, atm a cash withdrawal fee. A local_currency other than EUR means a foreign-currency payment.

4. Compute the run-rate. A line is recurring when the same counterparty appears in both windows, or is a direct debit, or is a known subscription. Normalise each recurring line to EUR per month from its billing cadence: a monthly charge counts once, even if it shows up twice in a window. Report the controllable run-rate per month and per year, and its change versus the previous window.