Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{201241,
        author = {Jayanthram K and Agilan ED and Gokul T and Harshavardhana Ram and Kishor G and Sakthivel S},
        title = {AI-Driven Document Intelligence based Supply Chain Automation Using Invoice Processing and Intelligent Querying},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {6872-6882},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=201241},
        abstract = {The rapid digitization of supply chain operations has intensified the need for intelligent systems capable of processing large volumes of semi-structured and unstructured documents such as invoices. Traditional methods relying on manual data entry or Optical Character Recognition (OCR) systems suffer from limitations in scalability, accuracy, and contextual understanding. This research presents an AI-driven document intelligence system designed to automate invoice processing and enable intelligent querying through a Retrieval-Augmented Generation (RAG) framework. The system integrates Vision-Language Models (VLMs), semantic embeddings, anomaly detection mechanisms, and knowledge distillation techniques to achieve both high performance and deployment efficiency.
A modular pipeline architecture is employed, consisting of an ingestion pipeline for document processing and a query pipeline for semantic retrieval and response generation. The ingestion pipeline performs preprocessing, hybrid extraction, structuring, validation, and embedding generation, while the query pipeline leverages vector similarity search and large language models to provide context-aware responses. Furthermore, knowledge distillation is applied to compress large language models into smaller, efficient student models, enabling faster inference and reduced computational cost without significant loss in performance.
Experimental evaluation demonstrates that the proposed system significantly outperforms traditional OCR-based approaches in terms of contextual understanding, flexibility across document formats, and query capabilities. The system achieves high accuracy in structured data extraction while maintaining scalability and usability for non-technical users. This work highlights the potential of combining document intelligence, semantic retrieval, and model compression to build practical, real-world AI systems for supply chain automation.},
        keywords = {},
        month = {May},
        }