Epapyrus Unveils Full Document AI Lineup at 2026 AI Summit

Epapyrus said on the 18th that it will participate in the '2026 AI Summit Seoul & EXPO,' to be held at COEX in Seoul. At this event, Epapyrus will showcase its document processing engine, 'PyMuPDF Pro,' and its web-based document solution, 'MuPDF WebViewer,' as its main offerings.

The event brings together global AI industry leaders and enterprise decision-makers. It is also a venue for discussing business applications of the latest generative AI technologies and innovations in AI infrastructure.

For companies and institutions pushing ahead with AI adoption, Epapyrus will unveil a high-performance unstructured document data pipeline, a factor that affects the success or failure of LLM and RAG systems. The unveiling will be accompanied by live demonstrations.

Through its participation in this event, Epapyrus aims to make inroads on site.

Epapyrus highlighted its differentiation point as a paradigm shift in document infrastructure in the AI era. As differentiating factors, it cited significant reductions in infrastructure costs compared with GPU server-centered document parser and chunking engines, as well as ultra-light deployment flexibility compared with GPU server-centered document parser and chunking engines.

To demonstrate this, it will unveil live demos of 'PyMuPDF Pro' and 'MuPDF WebViewer' at its booth. 'PyMuPDF Pro' supports complex PDFs, reports, and drawings, and enables ultra-fast extraction without losing text, layout, or image data. 'MuPDF WebViewer' is web-based, requires no separate installation, and supports immediate viewing of large documents. It also provides in-document Q&A and evidence-tracking features.

'PyMuPDF Pro' has surpassed 960 million global downloads and is backed by a proven performance base. It can also be built with just one lightweight Python package, and deployment environments include on-premises, closed networks, in-house PCs, and the cloud. In addition, 'PyMuPDF Pro' can complete data pipeline construction in 5 minutes.

The solution said it offers superior parsing and rendering speed compared with rivals and can operate independently of CPU/GPU specifications. It also focuses on reducing hallucination in LLM training by enabling precise parsing of complex table structures, multi-column layouts, and metadata within PDFs, Office files, and Hangul documents.

Along with this, 'MuPDF WebViewer' was also exhibited at the same time. 'MuPDF WebViewer' is a developer PDF web viewer SDK that can be immediately integrated with a single line of JavaScript code, and it was explained that it fundamentally blocks the risk of confidential document and personal data leaks by performing all processing only within memory inside the browser.

It also includes 'AI Citation' to support accurate tracing and highlighting of the original source locations behind AI answers, and it is also equipped with a 'sensitive information de-identification' feature. Search and annotation functions are also provided.

The company said existing similar solutions either impose heavy server deployment costs or remain limited to simple viewing. Kim Jeong-a, vice president at Epapyrus, said the company’s solutions make it possible to sharply reduce enterprise infrastructure deployment and operating costs and ensure the accuracy of LLM and RAG services, and added that AI Summit Seoul 2026 is an opportunity to demonstrate the differentiation of globally validated document AI technologies versus domestic competitors.

Epapyrus officially launched its next-generation web document viewer solution, 'StreamDocs 7.0,' on the 4th. The product is a next-generation offering that supports document viewing, security, and personal information de-identification at the same time.

Source: IT DAILY · Kwon Young-seok
Original: https://www.itdaily.kr/news/articleView.html?idxno=241041

References

• https://www.itdaily.kr/news/articleView.html?idxno=241041