Core features
Data integration
Get your numbers into a model: shaped templates, staged validation, and reviewable mapping.
Templates shaped to your model
Every model generates its own Excel template — the tables it actually needs, with validation rules baked in. Reach it from the Model Library row action Data integration — Excel template + validation rules, or from the Data requisition section of the build review, which lists "tables to provide · filled templates ready to download."
Filled templates are the fastest way to collect real numbers from the people who have them.
Staged, validated, reviewable
Uploads never apply directly:
- Stage & validate — files are staged and checked; findings come back before anything moves.
- Map data onto the verified structure — template-shaped uploads bind deterministically; unmatched tables can get AI mapping proposals that you review and correct. Fatal findings block the bind until fixed — or explicitly accepted as exceptions.
The uploader's contract: "quantities come from your data — relations are bound, never invented."
Sources today
- Excel workbooks (
.xlsx) — including full model imports through the reconciliation review. - Google Sheets — share the sheet "Anyone with the link → Viewer"; all tabs import.
- CSV — master data (
works.csv,products.csv,customers.csv) at authoring time, requisition CSVs at load time, and chat attachments. - Refresh from workbook — a model imported from Excel can re-sync when the workbook changes (the model row's refresh action).
Additional connectors are rolling out through the beta.