Agent-Driven Quality Assurance and Analytics with Gen AI and Graph RAG

Quality assurance is pivotal in driving operational excellence and enhancing export competence for MSMEs. At WCTE we have developed an Agent-Driven quality assurance and Analytics with Gen AI and Graph RAG and is being deployed at Shreyas Ltd. In this project over 300 historical non-conformance (NC) records were digitally transformed using prompt-driven Visual Document Understanding (VDU) models to rapidly convert legacy paper forms into structured digital data. After digitization, we compute similarity scores across individual fields to construct a knowledge graph— where each record is a node, and weighted edges capture similarity levels. Leveraging this robust framework, we are integrating an LLM-powered natural language interface via Graph Retrieval-Augmented Generation (Graph RAG) that enables agent-driven QA, supports autocompletion of NC records, and delivers detailed analytics on severity, frequency, and other key metrics for newly received forms, thereby fostering data- driven decision- making and bolstering operational

"Digitizing legacy NC records into a knowledge graph and layering it with an LLM interface—laying the groundwork for smarter QA and faster insights. This innovation streamlines quality assurance and fuels data-driven decision- making”

Dr. Ravi Hegde

Associate Professor, Dept. of Electrical Engineering, IIT-Madras

Key Features of the Platform

Legacy Digitization: Converts 300+ historical non-conformance records from paper into structured digital data using prompt-driven VDU models.

Knowledge Graph Construction: Computes similarity scores across fields to form a network where each record is a node connected by weighted similarity edges.

LLM-Powered Interface: Integrates Graph Retrieval-Augmented Generation (Graph RAG) for a natural language query interface.

Agent-Driven QA: Enables automated quality assurance with guided, data-driven actions.

Record Autocompletion: Supports predictive autocompletion for non-conformance records.

Detailed Analytics: Provides comprehensive insights on severity, frequency, and key metrics to drive operational excellence and export competence.