ClauseWise

Legal Clarity for Every Indian, Every Language

Comment

Inspiration

In India, millions unknowingly sign legal contracts, land deeds, and loan papers without fully understanding the fine print- legal jargons and language barriers keep them in the dark. Documents are often only in English or complex Hindi, and ordinary citizens, especially in rural and semi-urban areas, are left vulnerable to disputes and exploitation. What inspired us was seeing how confusing & complex legal documents are a struggle for many. ClauseWise is a tool aimed at bridging this language & knowledge gap, to make legal literacy a reality for every Indian, helping them make informed decisions without costly intermediaries.

What it does

ClauseWise is an AI-powered legal assistant that transforms complex legal documents into simple, local-language explanations. It leverages IBM’s open-source technologies—including the Granite series of models, and the Agent Development Kit (ADK)—to build an AI-powered tool that demystifies legal documents into plain, easy-to-understand language with support for multiple Indian languages. It highlights risks, obligations, and key crucial clauses, and enables a conversational Q&A where citizens can ask questions about their documents and receive clear answers, all in their preferred language.

How we plan to build it

  1. Document Ingestion – We begin with the feature to process scanned or digital PDFs of legal documents.
  2. Preprocessing – Using IBM’s OCR and data structuring toolkits, we clean and extract text from these documents.
  3. Granite Models – We will leverage Granite LLMs for clause classification, simplification into plain language, and translation into local Indian languages.
  4. Risk Highlighting – The system flags important obligations, risks, and critical clauses, present in the document, giving users a clear sense of what matters most.
  5. Interactive Q&A – Using the IBM ADK, we built a conversational agent that allows citizens to ask questions about their documents—in their mother tongue—and receive clear answers.
  6. Output – The final result: simplified, transparent, multilingual summaries that help individuals to understand legal documents, avoid exploitation and know exactly what they are entitled to.

Challenges we expect

  1. One of the biggest challenges is keeping the legal meaning intact while simplifying the language. Legal wording is very precise, and even small changes can alter the meaning.
  2. Another difficulty is ensuring that translations into Indian languages were accurate. India’s huge language diversity means ensuring accurate translations while also keeping it culturally sensitive, and easy to understand.
  3. India’s wide variety of legal contracts also poses a challenge, as our AI assistant needs to be able to handle any legal document.

Accomplishments that we're proud of

By focusing on real India-centric use cases like land deeds, rental agreements, and government schemes, we are proud to create a solution that will resonate with the everyday lives of millions in our country. It will be rewarding to see our project take a tough-looking legal clause and explain in their own language in a way they could immediately understand. This project gives us a chance to use AI for a social good and build an actual impactful project for the masses. We are also proud about the fact that we experimented with a new tech stack for this project and familiarized ourselves with new toolkits.

What we learned

Through this project, we explored IBM’s open-source ecosystem— worked with Granite models, OCR toolkits, and the Agent Development Kit (ADK)—and discovered how these tools can work together to build impactful AI solutions to real world problems. Working on this project we are learning how to optimize outputs for clarity and accuracy, while also keeping the user experience simple and transparent. We want to design a user-friendly interface, so ended up testing with different types of legal documents like agreements and contracts. The process was very iterative, we kept refining until the output felt right. So ultimately, we also gained a new knowledge about the complexity of legal structures in our country.

What's next for ClauseWise

  1. Expanding language coverage to include more Indian regional languages and dialects, ensuring inclusivity for maximum communities.
  2. Broadening document types to extend beyond just legal contracts to include medical consent forms, insurance policies and other critical citizen-facing documents.
  3. To build a lightweight mobile app so that users easily scan documents and receive instant simplified explanations.
Share this project:

Updates