Spreadsheet modeling

Inputs, calculations, outputs, checks

Your AI builds a spreadsheet model the way an analyst who gets audited would: assumptions on one sheet with sources, calculations identical across time, a single scenario switch, outputs that only reference, and checks that turn red.

By Vander Labs. A Vander Labs Original, written by the Vander Labs team with AI assistance. Works with Claude, ChatGPT, Cursor and Claude Code.

What your AI learns

• Separates inputs, calculations, outputs and checks into their own sheets
• Keeps formulas uniform across time with flags instead of one-off edits
• Adds a scenario switch, a sensitivity table and balance checks

How it works

1. Sketch the logic: The driver chain in words, each line becoming a row; anything not derivable becomes an input.
2. Lay out the workbook: Cover, Inputs, Calc, Outputs, Checks, Data and Notes, with one direction of flow.
3. Inputs sheet: One assumption per row with unit, source and Base, Low and High columns behind a single selector.
4. Consistent time: A single timeline row, period flags, identical formulas in every column, rollups by SUMIFS.
5. One step per row: Short formulas, no constants inside them, no volatile functions in large models.
6. Checks: Sources equal uses, reconciliations and bounds, each returning zero when fine, counted on the Cover.
7. Outputs: Five to eight decision numbers, a two-way sensitivity table and charts that only reference.
8. Harden and review: Validation, protection, version and change log, row-difference checks and extreme-input tests.

Limits and safety

• Modeling technique only; not financial, investment or tax advice, and says so when asked what the numbers mean.
• Refuses to type a number over a formula to make a period look right.

Topics: excel, google-sheets, forecasting, modeling, scenarios