Vakilo AI - Legal Assistant Platform

Abstract

Many people struggle to get legal help at the very start, because they cannot easily put their problem into legal terms or tell which lawyer would suit it. Vakilo AI is a web platform we built to guide that first stage: it walks a user through case intake and puts them in touch with verified legal professionals. The user describes the problem and chooses a category and an urgency level. An intake module powered by Gemini then poses ten follow-up questions and, from the answers, drafts a structured case summary with suggested next steps. The platform also proposes lawyers who may fit the case, and each of them can look over the request before accepting or declining it. Once a request is accepted, the two sides can chat privately, hold video consultations, book appointments, share case documents, and settle payments through Razorpay. A lawyer must pass administrative verification of their profile before being allowed to accept any case. The system is built from a Python FastAPI backend, a React front end, PostgreSQL and Supabase for data services, Socket.IO for messaging, and WebRTC for video. Vakilo AI is meant to bring order to the early stage of seeking legal help; what the AI generates is there to assist a qualified lawyer\'s advice, not to stand in for it.

Introduction

research paper presents Vakilo AI, an AI-assisted web platform designed to simplify access to legal services by connecting clients with suitable, verified lawyers through a single application. Finding legal assistance can be difficult because clients may not know which type of lawyer they need, what information to provide, or how to manage appointments, documents, communication, and payments. Traditional directories and referrals often do not provide all these services in one place.

1. Literature Review

Previous research has explored AI-based legal platforms, natural language processing (NLP), legal document analysis, case retrieval, and AI-generated legal answers. Although these technologies can improve access to legal information and organize legal workflows, challenges remain regarding accuracy, bias, explainability, and reliability. Research on Indian legal documents also indicates that AI systems may perform less accurately than human legal experts. Furthermore, retrieval-augmented generation (RAG) can help AI access relevant sources but does not eliminate the risk of generating incorrect information.

2. Objectives

The primary objective of Vakilo AI is to make legal assistance more accessible, organized, and convenient. Its specific objectives include:

• Providing a guided process for submitting legal cases without requiring prior legal knowledge.

• Using AI to ask relevant follow-up questions and generate structured case summaries.

• Recommending verified lawyers according to the client's case details and the lawyers' expertise.

• Enabling private messaging and video consultations between clients and lawyers.

• Integrating appointment scheduling, document sharing, payments, and notifications.

• Allowing administrators to verify lawyer profiles before they can accept cases.

3. Existing and Proposed Systems

Traditional systems rely mainly on personal referrals, online directories, and law-firm websites. These methods often lack integrated case management, lawyer verification, and communication tools, forcing clients to use multiple platforms.

In contrast, Vakilo AI combines the major stages of the legal assistance process within one web application. Clients submit their legal concerns, select a case category and urgency level, and answer ten AI-generated follow-up questions. The system then prepares a case summary, suggests possible next steps, and recommends lawyers from its verified directory. Lawyers can accept or decline requests, while administrators review lawyer profiles and supporting documents.

4. Methodology and Technologies

The system was developed using four main modules: client, lawyer, administrator, and AI-assisted intake. Its principal technologies include:

• Python and FastAPI: Backend development and REST APIs.

• React and TypeScript: Frontend interface.

• PostgreSQL and Supabase: Data storage and related services.

• Gemini API: Generating follow-up questions and case summaries.

• Socket.IO: Real-time messaging.

• WebRTC: Video consultations.

• Razorpay: Payment processing.

• Tailwind CSS and Vite: Interface styling and frontend development.

Separate dashboards are provided for clients, lawyers, and administrators to manage their respective activities and permissions.

5. Results and Discussion

The implemented prototype supports the main workflow from case submission to lawyer consultation. Its features include user registration, AI-assisted intake, case summaries, lawyer recommendations, administrator-led verification, private messaging, video consultations, appointment scheduling, document exchange, payments, and notifications.

The platform demonstrates the feasibility of combining these functions into a single application. Guided questions help clients organize their information, while case summaries may make consultations easier to prepare for. Integrated services also reduce the need to switch between separate communication and management tools.

However, the study does not report a controlled user evaluation, quantitative usability measurements, lawyer-matching accuracy, formal security testing, or independent assessment of AI-generated content. Therefore, the implementation demonstrates functional capabilities rather than proven improvements in efficiency, accuracy, or trustworthiness.

Conclusion

This paper described Vakilo AI, a web-based platform that organizes the early stages of seeking legal help. Users are guided through case submission, the Gemini API supplies follow-up questions, a structured summary is prepared, and lawyers whose profiles have passed administrator review are recommended. Messaging, video consultation, appointment scheduling, document exchange, payment processing, and real-time notifications are all brought into the same workflow. The implementation draws on FastAPI, React, TypeScript, PostgreSQL, Supabase, Socket.IO, WebRTC, and Razorpay. The project offers one practical way of linking AI-assisted intake with case coordination tools. Before stronger claims about real-world effectiveness can be made, however, the quality of the generated summaries and lawyer recommendations, along with the application\'s security, accessibility, and usability, must be evaluated systematically. Vakilo AI is meant to help users organize information and reach legal professionals, and it does not replace the judgment or advice of a qualified lawyer.

References

[1] Dr. Sudhakar Avareddy, Appu, Karthik M K, Mallikarjun, and Sai Shirisha, “Legal Assist: Smart Legal Case Management System,” International Journal of Engineering Research & Technology (IJERT), Vol. 14 | Issue 12 | December 2025, Paper ID: IJERTV14IS120677. [2] Deepanshu, Divi Saxena, Deepali Rana, Ayesha Varshney, and Sahinur Rahman Laskar, “NyayaAI: An AI-Powered Legal Assistant Using Multi-Agent Architecture and Retrieval-Augmented Generation,” arXiv preprint, Published: May 2026, arXiv:2605.10155, DOI: 10.48550/arXiv.2605.10155. [3] Amy J. Schmitz, Lola Akin Ojelabi, and John Zeleznikow, “Researching Online Dispute Resolution to Expand Access to Justice,” GiustiziaConsensuale (Consensual Justice), Vol. 1, No. 2, 2021, pp. 269–301, Ohio State Legal Studies Research Paper No. 680, DOI: 10.2139/ssrn.4018593. [4] Shubham Kumar Nigam, Shubham Kumar Mishra, Ayush Kumar Mishra, Noel Shallum, and Arnab Bhattacharya, “Legal Question-Answering in the Indian Context: Efficacy, Challenges, and Potential of Modern AI Models,” arXiv preprint, September 2023, arXiv:2309.14735. [5] T. Nava Krishna, B. Kusuma, K. Revanth, V. Reethi, K. Hemanth, and A. Prakash, “AI-Based Legal Information Retrieval Chatbot Using Natural Language Processing,” International Journal of Engineering Research & Technology (IJERT), Vol. 15 | Issue 3 | March 2026, Paper ID: IJERTV15IS030786. [6] Jhanvi Arora, Tanay Patankar, Alay Shah, and Shubham Joshi, “Artificial Intelligence as Legal Research Assistant,” FIRE Working Notes 2020, CEUR Workshop Proceedings, Vol. 2826, December 2020, pp. 60–65. [7] Yi Feng, Chuanyi Li, and Vincent Ng, “Legal Case Retrieval: A Survey of the State of the Art,” Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics, 2024, pp. 6472–6485. [8] Jorge Martinez-Gil, “A Survey on Legal Question–Answering Systems,” Computer Science Review, Vol. 48, May 2023, Article 100552, DOI: 10.1016/j.cosrev.2023.100552. [9] Rochak Bansal and Dr. Monica Chawla, “Transforming Legal Aid in India: Can AI Bridge the Justice Divide?” ShodhAI: Journal of Artificial Intelligence, Vol. 2 | Issue 1 | 2025, pp. 27–39, DOI: 10.29121/shodhai.v2.i1.2025.33. [10] Priyanka Prabhakar, Thanmai Gaddam, Divya Chennupalle, and Peeta Basa Pati, “Transformer-Based Embeddings and Stacking Classifiers for Judgment Outcome Prediction in Multilingual Legal Texts,” Scientific Reports, Volume 16, Article Number 18305, Published: 20 April 2026, DOI: 10.1038/s41598-026-47652-x. [11] Raihan Khan, Sayak Sarkar, Sainik Kumar Mahata, and Edwin Jose, “Security Threats in Agentic AI System,” arXiv preprint, October 2024, arXiv:2410.14728, DOI: 10.48550/arXiv.2410.14728. [12] Amy J. Schmitz, Lola Akin Ojelabi, and John Zeleznikow, “Researching Online Dispute Resolution to Expand Access to Justice,” GiustiziaConsensuale (Consensual Justice), Vol. 1, No. 2, 2021, pp. 269–301, Ohio State Legal Studies Research Paper No. 680, DOI: 10.2139/ssrn.4018593.

Copyright

Copyright © 2026 Yogesh Bisane, Roshni Sharma, Samanta Narde, Prachi Pandhurkar, Aumkar Sarode, Mr. Ashish Trivedi. 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.