evo.provensheet

Overview

Almost every company has one spreadsheet that quietly runs something that matters — pricing, estimating, quoting, commissions. It works, and it is usually better at its job than anything they could buy. It is also a liability: one person really understands it, it travels by email in versions, two people cannot work in it at once, and nothing stops a typed-in number replacing a formula.

ProvenSheet turns that workbook into a shared, multi-user web application — and proves the numbers survived. Upload the .xlsx and every formula is recomputed and compared against the result Excel itself cached in the file, so you are shown cell by cell that the maths is intact before anyone relies on it. Everything else in this space asks you to trust the conversion; the fidelity report is the product.

The proof case is the hardest workbook available: an industrial capital-project estimating system refined over 30 years of in-field use, pricing full manufacturing plant builds, factory overhauls, utility-scale wind farms, city power plants, and — at its most ambitious — the world's largest concrete plant (at the time of construction). It ships as ProvenSheet's built-in SWAG Estimates sample, and its 18 discipline-level tabs span civil, structural, mechanical, piping, electrical, production equipment and subcontractors. All 5,121 of its formulas match Excel's own results. Yours is probably simpler.

What the platform does is the same whichever workbook you give it: read the sheets, keep every formula live, and serve the result as a local-first web app. The sample is simply a demanding example of that — 18 discipline tabs, 1,453 input fields and 5,121 live formula cells that recalculate across sheets the moment any input changes. A workbook a tenth the size behaves the same way, and opens faster.

An added workbook becomes a template. Name a document and a project, and the platform clones that template — clearing the previous entries while preserving every formula, label, and cross-sheet dependency — so each new document starts from a fully wired copy rather than a blank page. Adding one is a drag-and-drop of an .xlsx or .xlsm file; the parser reads each cell's type, value, and number format and maps currencies, percentages, and dates to the appropriate UI inputs automatically.

Local-first architecture means people keep working where connectivity drops mid-session — a plant floor, a job trailer, a flight. Edits are written to the browser's own store first and pushed to the cloud database when the connection returns. The document list gives the whole workspace a live view by status — Draft, In Progress, or Completed — with open, clone, and delete controls.

What you get

• Any workbook, not a fixed schema — the application is generated from the sheets, formulas, and formats found in the file you add. The built-in sample is one workbook, not the product's shape.
• The fidelity report — every formula is recomputed and compared against the result Excel itself cached in the file, and the match is shown cell by cell before anyone relies on it. On the sample that is 5,121 of 5,121.
• Live cross-sheet formula engine — computed cells update in real time as inputs change, replicating the full Excel formula dependency graph in the browser via IronCalc (WASM).
• Template-based documents — an added workbook becomes the template; each new document clones it with prior data cleared and every formula, label, and dependency intact.
• Drag-and-drop import — drop in an .xlsx or .xlsm file; the parser identifies every cell's role (input, formula, label), maps currency and percentage formats to the right UI components, and converts dates to UTC for storage.
• Local-first operation — edits are written to IndexedDB first and queue there through a dropped connection, syncing to the cloud database automatically when it returns.
• Shared document list — every document with its project, edit count, status (Draft / In Progress / Completed), and who last changed it and when, with open, clone, and delete controls.
• A demanding sample, included — 18 discipline worksheets from a methodology proven on some of the largest industrial capital projects ever priced, open to explore without uploading anything.
• Flexible sidebar navigation — the worksheet menu can be docked left, right, or hidden, remembered per browser.
• Display preferences — timezone and date format (MM/DD/YYYY or DD/MM/YYYY) applied to every timestamp, remembered per browser.
• Multi-tenant cloud deployment — Docker + PostgreSQL architecture ready for AWS EC2; each client's data is isolated and the stack runs the same in development and production.

Inside the sample workbook

This is the structure of the built-in SWAG Estimates sample — an illustration of the kind of workbook the platform handles, not a layout it imposes. Your own file brings its own sheets. The sample's 18 discipline tabs match the breakdown used on real capital projects, and its Summary sheet rolls up totals from all of them in real time.

Technology stack

Application

Formula engine

Offline storage & sync

Data & persistence

Deployment & operations

The estimate data model uses a composite cell key (estimateId::worksheetName::cellAddress) so that cloud sync and offline mutations are fully idempotent — the same edit applied twice produces the same result.

Back to Work · Contact