Lomin is picking up pace in expanding its customer base across industries, led by its vision-language model (VLM)-based document automation business.

The company announced on the 28th that it recorded KRW 5 billion in cumulative bookings in the first half of this year. The result is attributed to wider adoption of its VLM-based Document AI Agent across the financial sector, along with an expanding footprint in the public, manufacturing, and healthcare sectors.

Since launching its Document AI Agent last year, Lomin has extended its business into document automation tailored to customers' own workflows. New orders have grown rapidly this year, centered on the financial sector. Three banks — including Woori Bank and Suhyup Bank — and two insurers, including Korean Re, have adopted projects built on the Lomin platform, and the scope of deployment is now widening to major domestic securities firms and capital finance companies.

Lomin is also broadening its customer base in manufacturing, the public sector, and healthcare, having secured customers such as Hyundai Motor Company, Hanwha Aerospace, the Korea Social Security Information Service, and GC Care.

The company points to its multimodal technology as the driver behind the order growth. VLM is a technology that interprets visual information and language together. Globally, document automation has been shifting quickly from a conventional optical character recognition (OCR) foundation toward multimodal AI capable of understanding context and document structure. Lomin is extending its application of VLM technology beyond simple information extraction into AI agents that understand documents and carry out the next task automatically.

Lomin also said it has consolidated its brand structure, unifying its former product brand, Textscope Studio, under the company name, Lomin.

The Lomin platform connects to customers' existing business systems through API integration. It automates the workflow from document upload to delivery of results, and provides document classification, key information extraction, document structure parsing, document comparison, de-identification, review, and training within a single workflow.