Give AI agents and ETL pipelines the spreadsheet calculations your business already runs.
Bring your workbook. Keep your formulas.
Runtime dependencies. Not a static graph.
evaltraCALCULATEPricing, commissions, allocations, forecasts.
Connect the model to the work around it.
Let your agent choose the scenario. Evaltra evaluates the workbook and returns the numbers, so the answer comes from your business model.
Run spreadsheet rules on fresh data inside a scheduled job. Feed calculated results into the next step, without opening the Excel application.
Inside an app, behind an API or in a notebook.
Your workflow owns the orchestration.
Dependencies follow the calculation as it runs.
Conditional formulas can use the path they actually need.
This SUMIF selects one row and leaves a circular reference unused. Evaltra and Excel calculate it. A static dependency check can reject it.
Explore the example ↗=SUMIF(B4:B5,1,C4:C5)| Cell / formula | EvaltraExample output | ExcelVerified reference | Other engines*Static dependencies |
|---|
SUMIF selects C4 = 7. The excluded row contains a reference back to the result, but it is never needed for this calculation.
Static preview: Evaltra shows prepared example outputs; Excel and the static engine show recorded reference results. *One tested static-dependency example, not every engine.
The Excel reference was recorded with iterative calculation disabled. This public preview uses prepared results and does not call the Evaltra engine.
Loading the example details…
Explore a scenario, a sales rule or a formula example.
Sample data, running right here in your browser.
Explore how pricing and customer growth shape a subscription business.
Apply spreadsheet rules to a sample dataset—the calculation step in a data pipeline.
Change A1, B1 and C1, then choose one of the four examples below. This preview supports these presets only.
Illustrative demo with synthetic data. No API connection.
Open a supported workbook, update its inputs and read the results. Keep the formulas your team understands.
import { Evaltra } from "@evaltra/engine";
const engine = Evaltra.browser();
try {
const book = await engine.open(workbookBytes);
await book.range("Inputs!B2")
.setCellContents([[59]]);
const result = await book
.range("Summary!B8").read();
console.log(result);
} finally {
await engine.close();
}Example for a workbook with Inputs and Summary sheets. Uses the locally built SDK package.
Your application maps a tool call or prepared dataset to model inputs, requests calculation and reads typed results and diagnostics. You control model access, orchestration and downstream actions.
A notebook, Databricks job or AWS Lambda function can be a client of the HTTP service where its runtime and network allow it. These are integration patterns; ready-made platform connectors and a native Python library are planned.
The native Rust executable exposes workbook sessions and isolated calculations through HTTP. The server does not require Node.js or Python at runtime. Keep the service token in your application backend; sessions and job results belong to one server process.
EVALTRA_SERVICE_TOKEN=… ./evaltra-server
POST /v1/sessions
POST /v1/sessions/{id}/commands
GET /v1/jobs/{id}Evaltra opens XLSX workbooks and runs supported formulas with the same Rust core in native and browser environments. Validate your model's formulas and workbook features before production use.
VBA and UDF execution, Power Query, automatic external-link refresh and iterative circular calculation are outside the runtime scope. Numeric and XLSX preservation compatibility has open work; successful import is not a guarantee that every formula or feature is supported.
This website is a static preview. Revenue and sales examples use simple browser calculations on synthetic data. The formula section implements four presets, and the comparison table displays prepared example outputs alongside recorded reference results.
No Evaltra API or engine runs on this page. The integration examples describe how you can use the product in your own application.