General Accounting Automation · Working Product
See the logic inside the workbook.
Business spreadsheets often run like undocumented software. Excel Intelligence & Controls maps each sheet's structure, groups repeated formula patterns, traces cross-sheet dependency evidence, preserves Excel errors, and saves a review queue locally.
Synthetic demonstration · Browser-local Step 0 · No production adoption claimed
Structure, formula behavior, review signals, and downloadable evidence remain connected to the source workbook.
What it helps you do
See how the workbook is built and where it may break.
The system does not replace accounting judgment or approve business rules. It shows the workbook's structure and control signals so a reviewer can inspect, test, and explain them.
Understand the workbook
Identify table shape, header depth, input and result columns, formula density, and the records each sheet appears to represent.
Map formula behavior
Group relative formula families, expose minority patterns, retain Excel errors, and summarize cross-sheet, external, or dynamic dependency signals.
Work from a review queue
Filter deterministic structural signals, inspect bounded evidence, and download matching Markdown, HTML, or JSON reports for another reviewer.
Working browser demo
Run it, then inspect the evidence.
The synthetic workbook contains payroll, inventory, department-master, and budget sheets. The same browser engine used by the Full Analyzer profiles all four.
- 1LoadRun the bundled workbook locally.
- 2ReviewSee the plain-English summary first.
- 3InspectOpen technical details or download evidence.
Synthetic accounting workbook4 sheets · seeded review issues · source labels projected to English
Ready. The workbook is processed only after you choose Run.
Start with the bundled synthetic workbook.
You will see what the analyzer found, why the signals matter, and where to inspect the underlying technical detail.
View complete technical results
What works now
The Browser Analyzer works here. The five Agent Skills run locally.
Both start with workbook evidence. Their runtime requirements are different.
This page profiles structure, formula families, dependency evidence, Excel errors, breakpoints, and review signals in browser memory. It needs no backend or model API key.
Build, Query, Standardize, Enterprise Management, and Ontology Git use Python and SQLite under an Agent host with human review. A public Web version of that workflow would need a secure backend and server-side credentials.
Analyze a workbook or inspect the saved test.
Use the Full Analyzer for your own workbook. Open Evidence for the saved synthetic run, system architecture, product guide, method, and build history.