NewsOur first product, Cellmark Editor, launched on October 22. Free for Windows.Learn more

Contact us
English한국어
Menu

Document digitizationTurn every document into accurate digital data.

Eureka-Vision · Document AI

Accurate digital data from any document.

Eureka-Vision turns titles, body text, tables, and formulas into data you can edit, search, and reuse. With local deployment on your PC or your organization's servers, documents stay within your environment.

Discuss API integration or technology transfer See it in Cellmark Editor
Demo of a PDF table being captured and converted into an editable table, preserving merged cells and header structure.
Drag to select and recognize a table in Cellmark Editor

Core technologies

Five technologies that turn documents into data

TitleBody textTableFormulaBody text

Document Layout Analysis

Identifies title, text, table, and formula regions so each can be processed appropriately.

1234512345

Reading Order Detection

Identifies the reading order of regions on pages with multiple columns, tables, and figures. It connects content in sequence for search, summarization, and use as AI input.

문서를 읽는 AI문서를 읽는 AIImageText

Korean OCR

Specialized OCR for Korean documents that converts text in paragraphs and tables into editable, searchable text.

Merged cellTwo-row header

Table Extraction

Offline · No external LLM

Recognizes table structure, including merged cells and multi-row headers, and represents it in LTEM, an independent open standard that anyone can use for free. Our own model runs offline on your PC, without calling external LLMs or sending documents outside your PC.

LTEM open standard

Where it runs

Process documents without sending them outside your organization

Eureka-Vision is light enough to run without high-end GPUs, so your documents don't have to go to outside AI services. Choose the setting that fits your rules.

On your PC

Runs offline, even on air-gapped networks

With Cellmark Editor, you can recognize and edit tables on your PC without an internet connection.

Within your organization

Runs on your own servers

Deploy within your organization to process documents without sending them outside.

Contact us

Eureka-Vision API · Cloud Available soon

For higher performance or larger document workloads

The Eureka-Vision API lets you integrate document layout analysis, reading order detection, Korean OCR, formula recognition, and table extraction into your product. Use it for higher performance or larger document workloads.

Discuss API integration

Our initial focus

For complex documents that need secure handling

Public administrationOfficial documents and forms with restrictions on external sharing
Education · Math textbooks and exam papersConvert math textbooks and exam papers, including formulas and diagrams, into accurate digital data. Exam papers that must remain confidential before release can be processed within your organization.Formula Recognition + Document Layout Analysis
Engineering drawingsWe are extending document recognition to fields such as ship design, where drawings hold essential information.

Products and technology

A product, an API, and an open standard

Product

Cellmark Editor

A free Markdown editor for Windows that uses Eureka-Vision for table recognition and LTEM to represent tables.

Explore Cellmark Editor

Open standard

LTEM

A free, open standard that extends Markdown to represent complex table structures, including multi-row headers, merged cells, and sections. The LTEM paper will be presented at EMNLP 2026 in Budapest.

Explore LTEM

API

Eureka-Vision API

The Eureka-Vision API lets you integrate document layout analysis, reading order detection, Korean OCR, formula recognition, and table extraction into your product. Use it for higher performance or larger document workloads.

Available soon