Compare the selected data.
Check selected tables, rows and cell values. Review changed values, missing or added data, and ambiguous matches. Confirmed column mapping and unique row keys help define the comparison.
SVERKA 2.0 / Created by ShaAnas
Check the transfer.
Then check the arithmetic.
A value can be copied correctly while the original calculation is wrong. SVERKA keeps these conclusions separate.
Research technical preview · Private pilots by arrangementThis public site explains the approach. It does not import documents, run the checking engine, or offer source or application downloads.
Synthetic illustration / Not a calculated result
Transferred value25.00 → 25.00
The selected value was copied unchanged.
Original calculation3 × 8.00 = 24.00
The written total of 25.00 differs from 24.00.
Two separate conclusionsTransfer ≠ arithmetic
A focused checking workflow
Find the difference.
Keep the scope explicit.
Review prepared tables, confirm what should be compared, and distinguish changed data from an incorrect source calculation.
Check declared rules separately.
Four supported rule types cover quantity × price, selected row sums, percentage shares totalling 100%, and matching values across tables. Units, period, scope and number format require review.
Keep findings with their context.
The private prototype supports prepared CSV/TSV tables and supported SVERKA JSON workspaces, approved arithmetic rules, settings profiles and separate reports. A workspace is not proof of an original document’s authenticity.
Review extracted table candidates.
A separate private local scanner prototype extracts candidate tables from supported digital PDFs with text. Experimental grid geometry supplements third-party PDF libraries. There is no OCR or automatic acceptance; each result needs human review.
See how the conclusions differ
Three readable examples.
Explore an unchanged value with incorrect arithmetic, swapped row values with the same grand total, and an incomplete comparison scope.
The walkthrough is manually prepared illustrative content. It does not run SVERKA or process files in your browser.
Open the case walkthroughFor a private pilot
- Describe the document types, expected checks and desired outcome.
- Agree a limited test with synthetic or suitably anonymised data and an appropriate processing environment.
- Review findings, uncertainties and possible improvements before a wider deployment.
Do not put confidential documents, credentials or sensitive personal data into a general enquiry.
Your documents / An agreed private test
Discuss a SVERKA pilot.
Tell us where table transfer or arithmetic checks take time. We can agree what to test and what evidence a useful result should include.
The public interface and illustrative example data are viewable and copyable. The checking engine and document processing are not included in these public pages.