Business administration · Handwriting and Form Recognition

OCR and handwriting recognition, Hungarian included, so paper becomes data

Has a complaint ever been lost because the signed delivery note could not be found?

Today, someone types up the signed delivery note, the handwritten report, the checklist, the form or the old register by hand days later, if it does not get lost in the meantime. The AI reads Hungarian handwriting too, and passes the content of each field as data to the ERP or the records system. Whatever it is unsure about is checked by a colleague, and the old archive becomes searchable.

In use within 4-8 weeks per document type · by scanner or mobile photo · on your own servers too

Does this sound familiar?

When the data is on paper and only turns up when there is trouble

This is what we hear from operations, back-office and document management leaders.

At the end of the month

The checklist is never evaluated

“I want to see in time if something doesn’t match, not at the end of the month.”

Head of operations, mid-sized company

At the audit

The data can only be checked by sampling

“A misread piece of data shouldn’t come to light at the audit.”

Head of document management, large enterprise

Retention rule

Retention cannot be checked for paper records

“I want to check the retention rule for the paper records too.”

Head of document management, large enterprise

The customer complains

Manual typing, mistyped data

“The customer shouldn’t complain because of a typo.”

Head of back office, large enterprise

Peak season

Cases pile up in peak periods

“When lots of cases arrive at once, the turnaround shouldn’t stretch to days.”

Head of back office, large enterprise

If someone leaves

Administration depends on one person

“If the administrator leaves, the data entry shouldn’t stop.”

Head of operations, mid-sized company

Not sure yet whether this is what you need?

Our expert will tell you whether it fits your situation.

What is at stake

If a delivery note is lost or illegible, the complaint cannot be closed.

The customer disputes the quantity, the carrier disputes the receipt, and the evidence is a sheet of paper somewhere in a file, or not even there any more. At that point it is not the administration they ask, but the head of operations, whose work only gets noticed when something goes wrong.

For each delivery note we record

the quantity, any discrepancy and whether it is signed, and the image is filed. If there is a dispute, nobody has to go through the files; a query in the system is enough.

How it works

From paper to the system, with human checks

01

It arrives

Scanned or photographed with a phone, from the field too.

02

The AI reads it

It recognises both printed and handwritten fields, and adds a confidence indicator to each.

03

What is uncertain, a person reviews

On the review screen, you only check what the machine is not sure about.

04

Data in the system

We pass the approved data to the ERP or the records system, and file the image of the document together with the data.

OCR, handwriting recognition, digitisation

Four document groups where paper costs the most

We start with one document type and measure how much it reads correctly before moving on to the next.

Logistics

Delivery notes and receipts

It reads the quantity, any discrepancy and whether there is a signature. Incoming Invoices and Receipts matches this against the invoice.

Field and plant

Reports and checklists

For handovers, maintenance or health and safety inspections. Instead of paper, they can also be recorded by voice: the voice note becomes a finished report and, if there is a discrepancy, a ticket too.

Administration

Forms

Application forms, trip sheets, questionnaires, declarations.

Archive

Old handwritten records

We make registers, machine logbooks and old contracts searchable, and you can ask questions about them.

What you get

Data from the paper, with the original image alongside

In a dispute or an audit, what you need turns up within minutes.

  • 01Field-by-field data from every document, with a confidence indicator
  • 02A review screen for uncertain fields
  • 03Image and data filed together
  • 04Full-text search and an archive you can ask questions about
  • 05A monthly report on accuracy and discrepancies
Before you decide

Worth knowing

Who it is for: operations, logistics and quality leaders, finance directors and records management leaders in transport, logistics, manufacturing, construction, utilities, field service providers, and any company with decades of paper archives.

What we need from you

A scanner or a mobile phone, and a few hundred samples per document type to measure accuracy and fine-tune.

Rollout

4-8 weeks per document type. Processing the old archive takes time depending on the volume. The price is set by the number of document types and pages.

What it builds on

Handwriting recognition AI, intelligent document processing, the DOCMAN document management system for filing, archive search, and the connection to the ERP and the Omnit business platform.

What it does not solve

It does not promise 100% accuracy, it does not decipher illegible writing, it does not verify the authenticity of signatures, and it does not replace the physical storage of paper.

The Handwriting and Form Recognition product sheet

What we did not cover on this page: the steps from a small first step to the complete solution, why it pays to go this way, and how we work along the way. Feel free to forward it to your colleagues.

Not sure whether paper is the place to start?

In 2-4 weeks, the Situation Assessment shows where most time goes on manual work and how many hours a week that adds up to. It costs HUF 199,000 + VAT, which we deduct from the price of your first development. If we do not find at least 20 hours a week of replaceable manual work, we refund the fee.

Frequently asked questions

What operations and document management leaders usually ask

We measure it per document type on a few hundred samples, and give you a monthly report on accuracy. We do not promise 100%, so whatever the AI is unsure about is checked by a colleague before it goes into the system.

An OCR program gives you text, while this reads the data field by field, with a confidence indicator and a review screen, and passes the approved data to the ERP or the records system.

No. It records whether there is a signature on the document, but does not verify its authenticity.

4-8 weeks for one document type. Processing the old archive depends on the volume.

For each document type, the price consists of a one-off set-up fee and a per-page fee; bulk processing of an old archive is priced individually. We give you an exact quote after the Situation Assessment or a conversation with one of our experts.

Also worth a look

Further reading

OCR in plain language

How character recognition works, what it does with handwriting, and where it is best run

The State of OCR Technology
Expert blog
The State of OCR Technology
OCR is not flawless. Understand realistic accuracy expectations, common limitations, and factors affecting performance in real-world applications.
The Complete Guide to Optical Character Recognition
Expert blog
The Complete Guide to Optical Character Recognition
Discover how OCR transforms scanned documents into usable data, reducing manual entry and streamlining document processing.
On-Premise or Cloud OCR
Expert blog
On-Premise or Cloud OCR?
On-premise or cloud OCR? Compare security, costs, performance, and data control to choose the right approach.
Just looking around?

Read on the blog about digitising paper-based processes, or subscribe to the Signal over Hype newsletter.

Business administration · Handwriting and Form Recognition

Interested in handwriting and form recognition? Let's talk.

Describe the situation in a sentence or two, and an expert in the field will get in touch with you.

Or call: +36 30 235 0100