This is the strongest part of the demo. You watch the software automatically separate a pile of mixed documents. It identifies an invoice (because it sees "Invoice" and a total due), separates it from a purchase order (PO number format), and isolates a W-9 form. No manual sorting. No drag-and-drop.
The "wow" factor comes from the . The demo almost always includes a scanned doctor’s prescription or a handwritten delivery note. ABBYY’s proprietary recognition engine (which is distinctly better than open-source Tesseract) usually nails the extraction, even with slanted cursive. abbyy flexicapture demo
Do not accept a generic demo. Demand a POC (Proof of Concept) using your vendor’s terrible invoices and your employee’s illegible handwriting. This is the strongest part of the demo
The danger of the ABBYY FlexiCapture demo is that it uses "optimized samples." The vendor chooses clean invoices and moderately messy handwriting. No manual sorting
The demo rarely shows the setup time. For FlexiCapture to recognize a "PO Number" on a specific vendor's form, a developer has to train the "Document Definition." While ABBYY has machine learning (ML) that reduces this time, complex layouts (tables inside tables, multi-page contracts with footnotes) still require significant manual zoning. The demo shows the result of training, not the process of training.