AllyOn NeoChat

The AI Chat Service is a professional service designed to help businesses move beyond the search inefficiencies and repetitive reporting burdens caused by data scattered across internal databases, documents, and unstructured sources, toward a single conversational interface for searching, analyzing, and acting on that data. This service features the construction of a RAG architecture that transforms internal repositories into a searchable knowledge base, natural-language-based querying of structured data, and insight generation through a Generative AI Chat powered by Amazon Bedrock—enabling companies to expand data access enterprise-wide, accelerate decision-making, and dramatically reduce repetitive manual work with minimal investment. Phase 1: Data Integration & Knowledge Base Construction In the initial phase, Doosan's experts work closely with the client to select the priority data sources for integration. Unstructured data such as internal documents, manuals, and bulletin boards is preprocessed and chunked, then embedded and stored as vector indexes on Amazon OpenSearch and Amazon S3 Vector, transforming it into a knowledge base capable of meaning-based search. At the same time, structured data stored in systems such as RDS and Redshift undergoes schema mapping, allowing the NL2SQL engine to learn the underlying table structures and accurately convert natural-language queries into SQL. The goal of this phase is to validate search accuracy and response quality within the client's actual data environment, and to lay the groundwork for scaling AI Chat adoption across the organization. Phase 2: Conversational Analysis & Value Proof Once data integration is complete, Doosan provides the key differentiator of the AI Chat Service: a Generative AI Chat powered by Amazon Bedrock. Without needing SQL expertise or data analysis skills, field and office staff can simply ask intuitive questions—such as, "Compare this month's equipment anomaly detections to last month's"—and receive answers that draw on both structured and unstructured data, along with supporting evidence and related documents. The service also includes data status analysis, automatic chart generation, historical data lookup, and report drafting, alongside PII detection, data leak prevention, guardrail options, and AI transparency reporting to ensure a secure environment for adoption. The goal of this phase is to give business users analyst-level access to data, ultimately accelerating data-driven decision-making and minimizing repetitive search and reporting work.