Software

Epapyrus Holds 2026 VIP Invitation Event

IT DAILY ·

ePapyrus’s "2026 VIP Invitation Event"现场 [Photo: ePapyrus]

✦ AI Summary

Epapyrus announced on the 15th that it had successfully held the '2026 Epapyrus VIP Invitation Event' at Lotte City Hotel Daejeon.

The event was organized for the company's VIP customers, including public institutions, government-funded research institutes, and university research labs, in line with the spread of AI Transformation (AX) centered on government ministries, local governments, public organizations, and research institutions.

Epapyrus introduced PyMuPDF Pro, MuPDF WebViewer, Native Extraction, StreamDocs AI, and StreamDocs 7.0, sharing technologies for building AI services for public institutions and research organizations and for addressing personal-information mishandling.

Epapyrus held the '2026 Epapyrus VIP Invitation Event.' Epapyrus announced on the 15th that the event had been successfully completed, and it was held at Lotte City Hotel Daejeon. The event targeted the company's VIP customers, including public institutions, government-funded research institutes, and university research labs, and it concluded successfully.

The event was organized in line with the spread of AI Transformation (AX) centered on government ministries, local governments, public organizations, and research institutions. Epapyrus prepared the event in response to the growing AX trend across the public and research sectors.

The main focus of the event was addressing a common pain point in the field: disappointing results after AI adoption. Epapyrus shared solutions and technical strategies for building trustworthy AI services for public institutions and research organizations.

The event also addressed the fact that companies and public institutions have identified AI hallucinations and mishandling of personal information as the biggest obstacles to adopting RAG and IDP. It also pointed out the limitations of existing AI document-processing methods, which focus only on simple text extraction, making it difficult to verify the source of AI outputs.

As a countermeasure, Epapyrus introduced its own technology, 'PyMuPDF Pro.' 'PyMuPDF Pro' can process HWP, Office, and PDF files, and it features the ability to extract high-quality layout data by understanding complex document structures and the reading order as seen by humans.

With integration with 'MuPDF WebViewer,' it became possible to highlight the source sentences behind AI-generated answers in real time on the original PDF. This allows users to quickly review results in an environment where accuracy can be verified within seconds.

Epapyrus then presented a technical alternative to address personal-information leakage issues in the handling of sensitive case records and public documents. It explained that standard personal-information masking, Blackmarking, appears as a black box on screen, but text data may remain inside the Blackmarking, and that residual text can be restored or extracted, creating serious security vulnerabilities in conventional Blackmarking.

In response, Epapyrus promoted an approach that combines 'Native Extraction' technology with AI contextual judgment. 'Native Extraction' is a technology that directly verifies and removes text at the original data level, and Epapyrus said its technology goes beyond simple OCR and visual-language-model (VLM) recognition by detecting and removing residual data behind black boxes, as well as hidden text inserted with transparent characters for the purpose of AI disruption, thereby fundamentally preventing mishandling of personal information. Epapyrus said it has presented a technology that combines direct verification and removal of original data with AI contextual judgment to address personal-information leakage in sensitive documents.

Epapyrus technology has been adopted in core infrastructure by global Big Tech companies such as Perplexity and OpenAI, and the maturity of Epapyrus technology was presented as already proven through these adoption cases.

On the day of the event, 'StreamDocs AI' drew attention. The purpose of 'StreamDocs AI' is to innovate resources in administrative and accounting workflows, and it is designed for approval and evidence-processing tasks at research institutions and companies. It helps reduce administrative delays caused by manual verification, reduce errors caused by manual verification, and shorten review time for spending requests. It includes streaming-based viewing without downloads, and, combined with AI analysis services, performs real-time rule checks, real-time automated summaries, and real-time error detection.

Epapyrus also officially launched the new product 'StreamDocs 7.0.' The product provides integrated support for all formats, including Hangeul, MS Office, and PDF, and offers powerful security options in a single solution, such as personal-information blackmarking and de-identification, invisible tracking codes, and time-stamp authentication (TSA) signatures. The event also included announcements of generous gifts for invited customers and special promotions for invited customers, drawing strong attention.

Kim Jeong-a, vice president at Epapyrus, said the key criteria for success or failure in AI adoption lie not in model performance or scoreboards, but in accurate reading in closed-network and high-volume processing environments, traceability of evidence, and the ability to maintain stable operations over time. She added that Epapyrus has 36 years of accumulated engine technology and document AI technology selected as infrastructure by global Big Tech companies. She also said the company aims to establish a global standard for a 'trustworthy AX without hallucinations or leakage' that public and enterprise customers can use with confidence.

Epapyrus also officially launched its next-generation web document viewer solution, 'StreamDocs 7.0,' in August. 'StreamDocs 7.0' supports document viewing, security, and personal-information de-identification features.

Source: IT DAILY · Kwon Young-seok
Original: https://www.itdaily.kr/news/articleView.html?idxno=241616

References

This article was produced with the help of an automated content generation algorithm.


Source: IT DAILY

View original

This article was summarized and organized by BizCrush based on the original article from IT DAILY. For exact quotations and full details, please refer to the original article.