Hand-filled forms are where data entry costs the most: every box someone wrote in has to be read and typed. SimpleIndex reads handwriting three ways, and you choose per field: ICR for printed characters in boxes, a cloud engine (AWS Textract, Microsoft Azure or Google) for unconstrained print and cursive, or a language model for the fields a template can't describe. Machine-printed text on the same page stays on the local OCR engine.
Handwriting isn't one problem with one answer. Every engine below reads it to some degree, and how well depends mostly on how neat the writing is. FineReader reads it better than Tesseract or SimpleOCR; Textract, Azure, Google and the AI models read it better still, at a per-page or per-token cost. You choose the engine for each field. Machine-printed text on the same page stays on the local OCR engine.
Reads hand-printed characters written in letter boxes or combs, the way most forms are designed, and it reads clear free handwriting too, less reliably the messier the writing gets. It reads handwriting better than Tesseract or SimpleOCR, runs on your machine, and has no per-page cost.
All three read handwritten text, including unconstrained handprint and cursive, along with labeled fields and tables. Textract reads cursive better than any local engine, though not perfectly; Azure Document Intelligence and Google Document AI suit organizations that already use them. The provider bills you per page.
Draw a zone around one handwritten box and hand just that zone to Claude, ChatGPT or Gemini, while the rest of the page stays on the local engine.
Checkboxes and filled circles come back as values, through mark recognition or a language model.
FineReader or Tesseract reads the machine-printed text on your hardware and makes the page searchable.
In job settings each field gets the engine it needs: ICR for boxed characters, a cloud engine or a language model for cursive.
And only on the pages you allow, so the rest of the batch costs nothing per page.
They can be checked against format rules or your own database before the batch is filed.
ICR stands for Intelligent Character Recognition: software reading hand-printed characters on a scanned page. Traditional forms constrain handwriting with letter boxes, so each character stands on its own, and ICR reads those well. For most forms processing that is the bulk of the work, and FineReader handles neat handwriting outside the boxes better than its reputation suggests.
Unconstrained handwriting is harder. In free-form comments, notes and cursive, letters join and run across the line. A cloud engine (AWS Textract, Microsoft Azure Document Intelligence or Google Document AI) or a language model reads these far better than a local engine, but not perfectly, so plan to check those fields. Flowing cursive prose, such as letters or long handwritten notes, is still where every method struggles.
Checkboxes and filled circles are read by mark recognition (OMR). Hand-filled forms are much of the work in healthcare, education and canvassing surveys. For how handwriting fits alongside the other recognition methods, see zone and dynamic OCR and every recognition feature. For job settings, see Handprint Recognition in the SimpleIndex wiki.
SimpleIndex Professional is licensed once, at $1,800. ICR through ABBYY FineReader runs on your machine with no per-page cost. AWS Textract, Azure Document Intelligence and Google Document AI are billed by their provider per page, and a cloud language model is billed by its provider by tokens. All of them are used only on the fields you send them.
SimpleIndex with AWS Textract reading handwritten characters.
Yes, with a cloud engine (AWS Textract, Microsoft Azure Document Intelligence or Google Document AI) or a language model, which read cursive far better than a local engine but not perfectly. FineReader ICR reads hand-printed characters, best in boxes, and neat handwriting outside them. Long flowing cursive, such as letters, is not reliable with any method yet.
Not for handprint: FineReader ICR runs on your own machine. Cursive and messy free-form handwriting read far better with a cloud engine (AWS, Azure or Google) or a cloud language model, and only the fields you choose are sent. If nothing may leave your network, version 12 can also send those fields to a language model on your own server; how well it reads cursive depends on the model you run, so test it on your own forms.
Professional, which includes ABBYY FineReader, AWS Textract and the AI engines. If you already own SimpleIndex OCR, its FineReader engine reads handprint too.
It depends on the handwriting and the form. Boxed handprint reads well; cursive reads better with a cloud engine but still needs checking. The free trial runs on your own documents, so you can measure it on the forms you actually receive.
No. SimpleIndex reads what's written. It doesn't compare handwriting samples to identify the writer.
The free trial runs the full software on your documents.
