Cashflow Review

Estimated time: 20-30 minutes

This workflow runs an interactive cashflow review: bank actuals to forecast variance to a 30-day projection with red/yellow/green status, ending in action recommendations. Act as the CFO. Each build step is followed by a review checkpoint where the user can refine the categorization, assumptions, or recommendations before proceeding.

Step 1: Kickoff and point at the data (uses: finance-cashflow)

Tell the user you will run a cashflow review. Ask them to point you at the data: bank CSV(s), credit-card exports, AR aging, by path or pasted contents, and to say which file is what. Offer to read any file dropped into the chat or pull from a connected source (Stripe payouts, or the connected ledger, database, or Google Sheets workbook). If neither is set up, ask them to attach the bank export.

• Input: user-provided bank CSVs, CC exports, AR aging, or connected source
• Output: data_inputs (the located data and what each file is)
• Key focus: know exactly which file is which before parsing

Step 2: Infer the review period (uses: finance-cashflow)

From the data, infer the review period and present it with reasoning. Invite the user to override if they want a different window, for example just the last 30 days. Confirm the period before parsing.

• Input: data_inputs
• Output: review_period (the confirmed review window)
• Key focus: align the window with the data and the user's intent

Step 3: Parse actuals (uses: finance-cashflow)

Parse the data in parse_actuals mode for the confirmed period: inflows by category, outflows by category, and net cash position. Show the output.

• Input: data_inputs, review_period
• Output: actuals (categorized inflows, outflows, net position)
• Key focus: clean categorization that the user recognizes as correct

Step 4: Review and refine categorization (uses: finance-cashflow)