RightsRoute: An AI-Powered Consumer Grievance Navigator with Document Intelligence and Retrieval-Augmented Generation
RightsRoute is an AI-assisted consumer grievance navigation system designed to help consumers organize complaints involving defective products, delayed refunds, warranty disputes, service deficiencies, and misleading commercial practices. The proposed system integrates natural-language complaint intake, document processing, optical character recognition (OCR), case structuring, timeline and evidence organization, consumer-information retrieval, documentation-readiness assessment, and editable complaint drafting. The application is implemented using React and TypeScript for the frontend and Python FastAPI for backend services. PDF and image processing is performed using PyMuPDF, Pillow, and Tesseract OCR. Consumer-information records are divided into overlapping text chunks of 850 characters with 120 characters of overlap and indexed using ChromaDB. Retrieval uses cosine similarity, with a default Top-K value of four chunks and a lexical keyword-overlap fallback. The system supports an optional OpenAI-compatible language-model API configuration for contextual response generation. The documentation-readiness score measures the completeness of provided case information and does not predict legal outcomes or the probability of grievance resolution. This manuscript describes the implemented prototype, its architecture, functional scope, testing approach, limitations, and future evaluation requirements. It does not claim quantified accuracy or validation on a representative real-world consumer grievance dataset. RightsRoute is intended as an assistive information-navigation tool and does not replace official consumer grievance authorities, legal advice, or professional legal representation.
Authors
- Chirayu Agrawal
- Rutuja Jabras
- aarya Bute
- vedant Bidwai
Institutions
- MIT Art, Design and Technology University (IN)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.5281/zenodo.24370195
- Primary Topic
- Artificial Intelligence in Law
- Type
- preprint