Extract data from cash flow statements

A cash flow statement answers a question the income statement cannot: where the cash actually moved. It reconciles the change in a company's cash balance across three sections, operating, investing, and financing, and an analyst reading it wants to see how much cash the business generated from operations before any accounting accruals. A lender testing debt service on a $6,000,000 facility, an equity analyst modeling free cash flow, and a treasurer forecasting liquidity all pull the same figures from a PDF: a statement out of an audited 10-K, an export from NetSuite, or a board pack rebuilt in Excel. The three section subtotals, net cash from operating activities, from investing, and from financing, sum to the net change in cash, which ties the opening balance to the closing balance for the period. The part that resists a flat read is the indirect-method reconciliation. Most statements start from net income and add back non-cash charges such as depreciation and amortization and stock-based compensation, then adjust for changes in working capital, before arriving at operating cash flow, so a line labeled Depreciation is an add-back here rather than an expense. Capital expenditures sit under investing as a cash outflow, while dividends paid and debt repayment sit under financing. Comparative statements place the current period beside the prior period, so a 2026-03-31 quarter sits next to 2025-03-31 and the columns must not blend. Parenthesized figures denote outflows. Talonic reads the cash flow statement and returns each section with its line items and subtotal, the reconciliation adjustments, and the net change that ties opening cash to closing cash, all tagged by period. A statement for the quarter ended 2026-03-31 reporting $920,000 of net operating cash flow in USD under US GAAP loads with its three sections intact, so an analyst verifies the cash bridge instead of retyping a page.

What gets extracted from cash flow statements

How extraction works for cash flow statements

Cash flow statements are pulled from audited filings, exported from NetSuite and Sage Intacct, and rebuilt in Excel, so the depth of each section and the labels on the reconciliation lines vary with the source and with the reporting standard, US GAAP or IFRS. Talonic reads the statement and maps it to the cash-flow schema in the Field Registry, which models the three activity sections and the net-change bridge rather than a flat list. Each line is tagged operating, investing, or financing and attached to its section subtotal, and the indirect-method reconciliation keeps net income, the non-cash add-backs such as depreciation and stock-based compensation, and the working-capital changes as distinct rows. The three subtotals are reconciled to the net change in cash, and the net change is checked against the opening and closing balances. Comparative columns are split by period. Parenthesized amounts are read as outflows. Every figure returns with a confidence score and pixel-region provenance under DIN SPEC 91491 conformity, so an analyst can verify an operating-cash-flow line against the source statement.

Sample extraction

A single-period cash flow statement prepared on the indirect method

{
  "entity_name": "Harbor Freight Components Inc.",
  "period_end_date": "2026-03-31",
  "fiscal_period": "Q1",
  "reporting_currency": "USD",
  "cash_flow_method": "indirect",
  "operating_activities": [
    {
      "line_item_description": "Net income",
      "current_period_amount": 352000
    },
    {
      "line_item_description": "Depreciation and amortization",
      "current_period_amount": 140000
    },
    {
      "line_item_description": "Change in accounts receivable",
      "current_period_amount": -68000
    }
  ],
  "operating_activities_net_cash": 920000,
  "investing_activities_net_cash": -310000,
  "financing_activities_net_cash": -150000,
  "net_change_in_cash": 460000,
  "cash_beginning_balance": 1650000,
  "cash_ending_balance": 2110000
}

Frequently asked