⚖️Levi- Legal AI Assistant

An AI-powered assistant for legal document intelligence. This project enables lawyers, paralegals, and businesses to upload legal documents and interact with them using advanced NLP features such as semantic Q&A, document verification, legal briefings, and clause extraction.

The system combines LLM-powered reasoning with custom document pipelines to help professionals save time, reduce errors, and get actionable insights from large legal texts.

Your multilingual, safe, and intelligent legal document companion — upload files, ask questions, or chat freely with general knowledge, all while ensuring no fabricated legal advice

---

🚀 Features

• 📂 Document Upload & Analysis – Upload legal files (PDF, DOCX, TXT) and run AI-powered verification, clause extraction, and compliance checks.
• 💬 Chat with Your Document – Ask document-specific questions in plain language; the AI retrieves context-aware answers.
• 🌍 Multi-Language Support – Ask questions in different languages and receive meaningful responses.
• 🤖 General Knowledge Mode – Even without uploading a document, you can ask general legal or non-legal questions, and the model will respond using its world knowledge.
• ⚖️ Safe by Design – The model never divises, invents, or fabricates legal advice; it sticks to document facts or general knowledge.
• 📝 Brief Mode – Generate concise summaries or structured briefings for faster case prep.
• 🔍 Document Verifier – AI-powered clause and compliance verification, returning structured outputs (JSON).
• ⚡ Smooth & Interactive UI – Modern Streamlit frontend for easy usage.

---

🏗️ Architecture

Streamlit UI
   |
   |--> File Loader (PDF/TXT/DOCX → Text)
   |--> Verifier Pipeline (run_document_verifier)
   |--> Brief Mode Pipeline (run_brief_mode)
   |--> Chat (ask_gemini with context + doc_text)
   |
   --> Backend Models (Google Gemini / FAISS index / Custom pipelines)

Key Modules:

• llm.py → ask_gemini(prompt, document) handles context-aware legal Q&A.
• verifier.py → run_document_verifier(text) runs compliance + legal checks.
• briefings.py → run_brief_mode(mode, text) generates concise legal briefings.
• utils/file_loader.py → Handles PDF, DOCX, and TXT ingestion.

---

⚙️ Installation

Clone the repository:

git clone https://github.com/your-username/legal-ai-assistant.git
cd legal-ai-assistant

Create and activate a virtual environment:

python -m venv .venv
.venv\Scripts\activate   # On Windows
source .venv/bin/activate  # On macOS/Linux

Install dependencies:

pip install -r requirements.txt

---

▶️ Running the App

To launch the Streamlit UI:

streamlit run app.py

Open in browser at [https://agent-translator-mgtwfgmztybxyfkw9pbddx.streamlit.app/]

For more Demo and Screenshots

[https://drive.google.com/drive/folders/1lMOVf16aaa84_eu4Uv-0zCsvFVwEC49W?usp=sharing]

📌 Usage

1. Upload a document (PDF, DOCX, TXT) from the sidebar.

2. Choose an action:

   • 🔍 Run Verifier → Get structured JSON results of compliance checks.
   • 📝 Generate Briefing → AI-generated legal summary.
   • 💬 Chat → Ask questions directly about the document.

3. Review AI outputs, copy insights, or export JSON results.

---

🎥 Demo

1. Upload Document

---

2. Run Verifier

---

3. Generate Briefing

---

4. Chat with Document

---

🎬 Full Demo Folder

👉 View All Demos

---

🛠️ Tech Stack

• Frontend: Streamlit
• Backend: Python (FastAPI-ready endpoints)
• LLM Integration: Google Gemini API
• Embeddings / Indexing: FAISS
• Utilities: PyPDF2, python-docx, JSON, Pandas

---

📂 Project Structure

legal-ai-assistant/
│── app.py                 # Streamlit frontend
│── llm.py                 # LLM integration (Gemini, embeddings)
│── verifier.py            # Document verification pipeline
│── briefings.py           # Legal briefing pipeline
│── utils/
│   └── file_loader.py     # Document ingestion utilities
│── data/
│   └── faiss_index.bin    # Vector index for semantic search
│── assets/
│   ├── screenshots/       # Static UI screenshots
│   └── demo/              # Demo GIFs
│── requirements.txt
│── README.md

---

🔮 Roadmap

• Export results (JSON/PDF briefings).
• Add clause comparison across multiple documents.
• Integrate legal precedent search with multi-doc context.
• Deploy as a FastAPI backend + Streamlit frontend for production use.
• User authentication & document history.

---

🤝 Contributing

Contributions are welcome! Please open an issue or submit a pull request.

---

📜 License

This project is licensed under the MIT License.

---

⚖ Built with AI to make legal work simpler, faster, and smarter.

---