Document AI company Lomin said on Tuesday it supplied its document AI agent, 'Lomin', to Korean Reinsurance’s 'special insurance rate calculation AI assistant 구축 project'.

The project applies AI to rate calculation work in the special insurance underwriting process.

In the project, Lomin was responsible for retrieval-augmented generation (RAG) preprocessing for the rate calculation AI assistant. It classified and parsed submitted documents such as rate request forms and questionnaires, extracted necessary information and converted it into structured data.

For rate request forms written across multiple pages, it developed and applied a multi-page document extraction AI model.

'Lomin' is a document AI agent based on a vision-language model (VLM) that processes various unstructured documents and automates document tasks by linking with corporate business systems. It provides a grounding function that allows users to 확인 the location of evidence for processing results in the original document, and offers field-by-field confidence levels.

Lomin Chief Executive Ji-hong Kang (강지홍) said, "To apply generative AI and AI agents to real work in finance, including insurance, the key is accurately turning complex and diverse documents into data." He said, "Based on our experience performing document datafication in large-scale AI projects for major insurers and banks and securities firms, we will lead the document AI field for generative AI and AI agents in the financial sector."