Structured Legal Document Generation in India: A Model-Agnostic Wrapper Approach with VidhikDastaavej

TL;DR

The paper tackles the challenge of automating private legal document drafting in India by introducing the anonymized VidhikDastaavej dataset and a domain-adapted model, NyayaShilp. It presents a Model-Agnostic Wrapper (MAW) that enforces a two-phase, retrieval-augmented generation process to improve coherence, factual accuracy, and reduce hallucinations in long-form documents. Experimental evidence shows that while direct fine-tuning on a small private dataset yields limited gains, the MAW wrapper enables open-source models to achieve performance approaching GPT-4o on structured drafting tasks, with strong expert validation. The work also provides an interactive HITL system to support real-world legal drafting workflows and discusses ethical considerations, privacy, and future directions for broader deployment and cross-jurisdiction adaptation.

Abstract

Automating legal document drafting can significantly enhance efficiency, reduce manual effort, and streamline legal workflows. While prior research has explored tasks such as judgment prediction and case summarization, the structured generation of private legal documents in the Indian legal domain remains largely unaddressed. To bridge this gap, we introduce VidhikDastaavej, a novel, anonymized dataset of private legal documents, and develop NyayaShilp, a fine-tuned legal document generation model specifically adapted to Indian legal texts. We propose a Model-Agnostic Wrapper (MAW), a two-step framework that first generates structured section titles and then iteratively produces content while leveraging retrieval-based mechanisms to ensure coherence and factual accuracy. We benchmark multiple open-source LLMs, including instruction-tuned and domain-adapted versions, alongside proprietary models for comparison. Our findings indicate that while direct fine-tuning on small datasets does not always yield improvements, our structured wrapper significantly enhances coherence, factual adherence, and overall document quality while mitigating hallucinations. To ensure real-world applicability, we developed a Human-in-the-Loop (HITL) Document Generation System, an interactive user interface that enables users to specify document types, refine section details, and generate structured legal drafts. This tool allows legal professionals and researchers to generate, validate, and refine AI-generated legal documents efficiently. Extensive evaluations, including expert assessments, confirm that our framework achieves high reliability in structured legal drafting. This research establishes a scalable and adaptable foundation for AI-assisted legal drafting in India, offering an effective approach to structured legal document generation.

Figures (5)

• Figure 1: Wrapper flow diagram
• Figure 2: Document Information Entry Interface
• Figure 4: Adding Document Sections
• Figure 5: Generating Section Content
• Figure 6: Exporting the Document