🏛️ Courtroom-3D

📋 Overview

Courtroom-3D is an interactive legal advisor application that provides concise legal information through an AI-powered interface. The system uses Google's Gemini AI to generate legal advice in response to user queries.

✨ Features

• 🤖 AI Legal Advisor: Get legal insights from an AI trained to respond like an experienced legal professional
• ⚡ Concise Responses: All legal advice is limited to 60 words for quick, actionable information
• 🔌 API Integration: Uses Google's Gemini 2.0 Flash model for generating responses

🛠️ Technical Stack

• Frontend: Next.js
• Backend: Next.js
• AI: Google Gemini API
• Language: TypeScript

🚀 Getting Started

📥 Clone the repository:

git clone https://github.com/yourusername/courtroom-3d.git
cd courtroom-3d

📦 Install dependencies:

pnpm install

🔑 Create a .env.local file with your API key:

GEMINI_API_KEY
ELEVENLABS_API_KEY

🎯 Start the development server:

pnpm dev

💡 Usage

1. 📝 Enter your legal question in the provided interface
2. 📨 Receive a concise (60 words or less) response from the AI legal advisor
3. 🎯 Use the information as a starting point for addressing your legal concerns

🎯 Our Core Contribution: Building the Blueprint

We are defining the robust, user-centric, and privacy-first technical framework for the entire legal assistant. This ensures we have a clear roadmap for expanding beyond initial features, maintaining technical integrity, and delivering maximum impact.

✨ Key Architectural Pillars (How We're Proceeding)

We're building this project on three main pillars:

1. Comprehensive Legal Education System:

   • How: This involves setting up pipelines to ingest and process vast legal content (books, lectures, PYQs) in multiple regional languages. We will develop an interactive AI avatar tutor (using TTS/STT and an LLM) to deliver engaging, simplified legal lessons.
   • Why: To democratize legal literacy and make complex laws easy to understand for everyone, reducing reliance on expensive traditional methods.

2. Private AI Legal Advisor:

   • How: We're designing a secure backend (FastAPI) where an LLM (running locally via Ollama or via secure APIs like Gemini, as used in frontend) retrieves context from a specialized legal vector database. This powers concise legal advice based on verified documents and court judgments.
   • Why: To provide immediate, basic legal guidance privately and affordably, addressing the high cost and fear associated with traditional legal consultations.

3. Robust Privacy-by-Design Layer:

   • How: This is integrated at every level, ensuring zero data collection and no server-based logging of user queries. Focus is on local-first processing and transparent open-source components.
   • Why: To build absolute trust with users, especially given the sensitive nature of legal queries, ensuring their data ownership and confidentiality.

🛠️ Technical Approach (Under the Hood)

Our architecture leverages open-source AI and data tools for flexibility and cost-effectiveness:

• LLMs: Mistral 7B / Phi-3 / Gemma (for local/private processing) and integration with models like Google Gemini (as used in frontend).
• Retrieval-Augmented Generation (RAG): Using Langchain/LlamaIndex with Vector Databases (ChromaDB/FAISS) for accurate, contextual responses.
• Speech & UI: Integrating Whisper (STT), Coqui/Tortoise/ElevenLabs (TTS), and Streamlit/React for user-friendly interfaces.

🚀 Moving Forward

This architectural blueprint guides our development, ensuring that Courtroom-3D can systematically evolve into the full ApnaWakeel.ai vision, expanding its features (like document analysis, hearing trackers, template generation), increasing language support, and scaling effectively while always prioritizing user privacy and impact.

Architecture