From every document, the next eureka.

We build AI that turns documents into data and accelerates scientific discovery.

Document digitization

Turn every document into accurate digital data.

Product family

Eureka-Vision · Document AI

Our table recognition technology powers Cellmark Editor, a free Markdown editor for Windows.

Document Layout AnalysisIdentifies and locates titles, body text, tables, and formulas on a page.
Reading Order DetectionIdentifies the reading order of pages with multiple columns, tables, and figures.
Korean OCRBuilt for Korean documents. Converts Korean text in images into editable text.
Formula RecognitionConverts formulas in images into editable math.
Table Extraction + LTEMOffline · No external LLMPreserves table structure, including merged cells and multi-row headers, in LTEM.

Science Acceleration

Understanding how science advances. Bringing the next eureka closer.

Product family

Eureka-Scholar · Research AI

Starting with natural language processing and machine learning.

IdeaForestExplore how research fields emerge, connect, and evolve.Available soon
Paper ReaderQuickly grasp the key ideas in a research paper.Available soon
Paper FigureCreate charts that clearly convey the key message of your paper draft.Available soon

Product

Cellmark Editor

A free Markdown editor for Windows with support for merged cells and multi-row headers.

Download for WindowsVersion 0.1.2 · 125 MB · Version list Explore Cellmark Editor

Free for everyone Free for individuals, businesses, and public organizations, including use at work.
Your documents never leave your PC.
Table recognition by Eureka-Vision. Table representation in LTEM. Capture a table on screen, and offline AI on your PC turns it into an editable table.

Cellmark Editor showing the same table with merged cells and two header rows in a PDF on the left, Markdown in the center, and Publish view on the right. — A sample document in Cellmark Editor

Our direction

Digitize documents. Connect knowledge. Run AI anywhere.

01Now

Digitize

We build technology that turns records in many forms into digital data, starting with documents.

Eureka-Vision
02Next

Structure

We structure and connect data to build usable knowledge, such as mapping how AI research has evolved.

LTEM · Eureka-Scholar
03Long term

Run intelligence anywhere

We are working toward capable, lightweight AI for environments with security and resource constraints.

Ultra-light language models

Open standard

LTEM: an open standard for complex tables in Markdown.

LTEM is an independent open standard developed by EurekaAI, free for anyone to use. It represents complex table structures—including multi-row headers, merged cells, and sections—in Markdown that people and AI can read. The LTEM paper will be presented at EMNLP 2026 (Budapest, October 24–29).

Explore LTEM Official LTEM website

LTEM source and table preview. Revenue spans the Q1 and Q2 columns, while Region spans both header rows.

Three ways to use our technology

Developers · Eureka-Vision API Available soon

One API for document recognition

The Eureka-Vision API lets you integrate document layout analysis, reading order detection, Korean OCR, formula recognition, and table extraction into your product.

Discuss API integration

Businesses and public organizations

Process documents without sending them outside your organization

For environments where data must stay within the organization, we can discuss internal deployment, fully offline operation, and technology transfer of our core engines.

Research · Open standards

Explore our research and open standards

Use LTEM, our open standard for tables, and follow the research behind Eureka-Scholar.

Explore

LTEM

News