KR EN

LLOYDK products built around your enterprise knowledge and workflows

DO-RAG

DO-RAG

Turn enterprise documents into trusted knowledge


DO Solution 2.0 DO-RAG is an enterprise RAG platform that converts enterprise documents into knowledge AI can understand and answers questions with evidence tied to exact pages and locations in the original documents.

  • Knowledge that preserves document structure

    Identifies pages, headings, paragraphs, tables and their rows and columns, figures, formulas, and original locations as individual units of knowledge.

  • Search by meaning and exact terminology

    Combines semantic and keyword search, then reranks results to reduce omissions and errors.

  • Answers backed by precise evidence

    Provides the pages, sections, tables, and locations in the source documents used to generate each answer.

  • Hybrid Search

    Semantic search + keyword search

  • Reranking

    Prioritize the most relevant results

  • Evidence-Based Answer

    Show evidence alongside answers

Search Quality & Operations

Search quality and operations built for enterprise environments

DO-RAG preserves the structure and meaning of enterprise documents rather than flattening them into plain text.
It uses multiple search methods to find the information you need and provides clear source references.

01

Data-driven search quality management

Continuously validates search results to improve accuracy.

02

Operations tailored to your organization

Systematically manages document spaces and user permissions for each team.

Beyond General RAG

How does DO-RAG differ from conventional RAG?

From document processing and search to supporting evidence and quality management,
DO-RAG is designed to help organizations use enterprise documents more accurately and systematically.

  • Table row and column structure is lost

    Preserve table structure and headings

  • Vector search

    Vector, keyword, and hybrid search with reranking

  • No evidence, or only a file name

    Source references down to the page, section, table, and location

FAQ

Frequently asked questions about our products

Find answers to common questions about deployment and use.

  • Q1

    Can it process PDFs, HWP files, and scanned images?

    A
    DO-RAG supports PDF, HWP, HWPX, Word, Excel, PowerPoint, images, and scanned documents.
     It applies OCR to scanned documents, allowing you to upload and process originals without converting them to another format.
  • Q2

    표·그림·수식이 포함된 문서도 원본 구조를 유지해 검색할 수 있나요?

    A
    DO-RAG는 문서를 단순 텍스트로 변환하지 않고 제목·문단·표·그림·수식·페이지와 원문 위치를 보존해 검색 가능한 지식으로 전환합니다.
    의미 검색과 키워드 검색을 결합한 Hybrid Search와 Reranking을 적용해 문서의 구조와 정확한 업무 용어를 함께 검색할 수 있습니다.
  • Q3

    AI 답변의 출처와 검색 품질은 어떻게 확인할 수 있나요?

    A
    DO-RAG는 답변에 활용된 문서명과 페이지, 섹션·표 및 원문 위치를 출처로 제공합니다.
    QA Dataset과 MRR·Recall@K 등의 지표를 활용해 검색 설정별 결과를 비교하고 검색 품질을 지속적으로 관리할 수 있습니다.

Contact Us