LegalEase: An AI-Driven Assistant for Legal Query Resolution and Document Summarization in India

Abstract

Due to complicated legal jargon, unreliable documentation systems, and a general lack of knowledge of procedural requirements, ordinary persons in India still have very limited access to legal information. Because of this, people frequently rely extensively on legal experts for even the most basic comprehension, which creates substantial cost and informational hurdles. Large language models (LLMs) and other recent developments in artificial intelligence have made it possible to provide simple, scalable, and reasonably priced legal aid. This study provides a full AI-driven legal assistant specifically built for Indian law that supports statute-level information retrieval, natural-language querying, and legal document summarization. The system exhibits an effective balance between accuracy, speed, and accessibility. It is built with a React frontend, Node.js backend, MongoDB storage, and GPT-4o-mini. We assess the system using ten real-world legal papers from the criminal, civil, procedural, and family law sectors, as well as fifty different legal queries. The assistant performed well in terms of clarity, summary quality, and user pleasure, with an average accuracy of 82%. The system architecture, data processing methodology, prompt engineering strategies, limitations, ethical considerations, and future improvements like multilingual support, retrieval-augmented generation (RAG), and domain-specific fine-tuning are all described in this paper in addition to the results. The results show that lightweight LLM-based solutions have a great deal of promise to empower individuals, students, and professionals looking for easily available legal clarity and to democratize legal knowledge.

Introduction

India’s legal system is vast and complex, making legal understanding difficult for most citizens due to complicated language, fragmented information, high consultation costs, and limited legal education. Although digital legal databases exist, they rely on keyword searches and legal expertise, while most AI-based legal tools are designed for Western jurisdictions and do not adapt well to Indian laws. This creates a strong need for an accessible, India-specific legal assistant.

The paper introduces a lightweight Legal AI Assistant designed to help non-lawyers with law lookup, document summarization, and natural-language legal guidance. Rather than focusing on judgment prediction or deep legal reasoning, the system emphasizes usability, affordability, and practical legal support through efficient workflows and large language model prompting.

The literature review shows that advances in NLP and transformer-based models have improved legal document processing, semantic search, and contract analysis. However, gaps remain in Indian legal tools, including limited AI-driven summarization, overreliance on keyword search, lack of Indian datasets, high costs, and the absence of an integrated platform for document upload, analysis, and case retrieval.

Related work highlights that existing AI systems largely target Western legal frameworks, focus on computationally heavy judgment prediction, or address narrow legal domains. The problem statement identifies key challenges in India’s legal ecosystem: complex legal language, poor accessibility, high costs, fragmented sources, and the need for simplified guidance.

Conclusion

A useful, approachable, and effective AI-powered legal assistant designed for Indian legal situations was described in this study. The technology greatly increases legal accessibility and awareness for general users through conversational engagement, document summarizing, and legal retrieval. According to experimental results, the assistant performs well in terms of correctness and clarity, making it appropriate for usage in educational and informative settings. The technology significantly lowers informational barriers and improves public understanding, even though it is not meant to replace legal specialists. Future improvements, such as RAG integration, fine-tuning, multilingual capabilities, and reference checking, might make the assistant a highly influential instrument in India\'s judicial system.

References

[1] Rabelo, J., Naique, J., Chalkidis, I., & Neves, L. (2023). LexGLUE: A benchmark dataset for legal language understanding in English. Findings of the Association for Computational Linguistics. [2] Henderson, P., et al. (2023). Pile of Law: Learning responsible data filtering in legal NLP. In Advances in Neural Information Processing Systems (NeurIPS). [3] Zhong, H., Xiao, C., Tu, C., Liu, Z., & Sun, M. (2022). Legal judgment prediction via legal reading comprehension. Artificial Intelligence and Law. [4] Long, Y., Chen, C., Liu, Y., & Ye, Q. (2022). CaseFormer: Pre-trained transformer for legal case retrieval. Findings of the Association for Computational Linguistics. [5] Zeng, J., Lei, Q., & Chen, Z. (2022). Explainable legal judgment prediction using hierarchical attention networks. Knowledge-Based Systems. [6] Xu, W., Li, Z., & Wang, P. (2021). Automatic contract element extraction using deep learning. Artificial Intelligence and Law. [7] Hendrycks, D., et al. (2021). Measuring massive multitask language understanding (MMLU). International Conference on Learning Representations (ICLR). [8] Chalkidis, I., Fergadiotis, M., Malakasiotis, P., & Androutsopoulos, I. (2021). Paragraph-level rationale extraction in legal texts. Proceedings of the Association for Computational Linguistics (ACL). [9] Li, J., Tang, X., Li, L., & Barton, T. (2021). CUAD: The Contract Understanding Atticus Dataset. arXiv. https://arxiv.org/abs/2103.06268 [10] Chalkidis, I., Fergadiotis, M., Tsarapatsanis, D., Androutsopoulos, I., & Aletras, N. (2020). LEGAL-BERT: The Muppets straight out of law school. arXiv. [11] Zhong, H., Guo, Z., Tu, C., Xiao, C., Liu, Z., & Sun, M. (2020). JEC-QA: A legal-domain question answering dataset. Proceedings of the AAAI Conference on Artificial Intelligence. [12] Xu, W., Li, C., Lee, M., & Zhang, C. (2020). Multi-task learning for abstractive text summarization with key information guide network. EURASIP Journal on Advances in Signal Processing, 2020(1). [13] Chalkidis, I., Malakasiotis, P., & Androutsopoulos, I. (2020). Legal judgment prediction for the European Court of Human Rights. Artificial Intelligence and Law. [14] Grabmair, M., et al. (2019). Introducing LUIMA: A legal argument mining framework. Artificial Intelligence and Law. [15] Sugathadasa, S., Ranasinghe, T., & de Silva, N. (2019). Legal document retrieval using deep neural models. Proceedings of ICAART. [16] Nguyen, D. Q., & Nguyen, A. T. (2018). A legal case retrieval system using deep Siamese networks. Proceedings of the 27th International Conference on Computational Linguistics (COLING). [17] Kaur, P., & Chatterjee, A. (2018). Automatic legal document summarization using machine learning. International Journal of Computational Linguistics. [18] Katz, D. M., Bommarito, M., & Blackman, J. (2017). A general approach for predicting the behavior of the Supreme Court of the United States. PLoS ONE, 12(4), e0174698. [19] Sulea, O.-M., Zampieri, M., Malmasi, S., & van Genabith, J. (2017). Exploring the predictability of judicial decisions in the French Supreme Court. arXiv. [20] Aletras, N., Tsarapatsanis, D., Preotiuc-Pietro, D., & Lampos, V. (2016). Predicting judicial decisions of the European Court of Human Rights: A natural language processing perspective. PeerJ Computer Science, 2, Article e93. [21] Kumar, S., Jha, S., & Garg, A. (2024). NyayaAnumana: A large-scale Indian legal judgment prediction dataset. arXiv.

Copyright

Copyright © 2025 Priyanshu Srivastava, Rohit Singh, Rishabh Agrawal, Utkarsh , Mr. Abhishek Shahi. This 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.