Samvaad (संवाद)

Dialogue-Based Learning Facilitated by Real-Time AI

Samvaad is a sophisticated, full-stack Retrieval-Augmented Generation (RAG) platform that transforms static documents into interactive, dialogue-driven learning experiences. By combining state-of-the-art vector retrieval with ultra-low-latency WebRTC voice conversations, Samvaad allows you to talk to your knowledge—literally.

Key Features

• Real-Time Voice Mode: Ultra-low latency voice conversations using WebRTC (Daily) and the Pipecat framework. Interact naturally with your documents using voice.
• Deep RAG Pipeline: Advanced document ingestion supporting PDF, Docx, Images (OCR), and more via LlamaParse and Docling.
• Intelligent Deduplication: Content-addressable storage (SHA-256) ensures zero-redundancy in document chunks across users.
• Multi-Model Intelligence: Optimized for Gemini, Groq, and OpenAI, with high-quality embeddings via Voyage AI.
• Neural TTS & STT: Crystal-clear speech synthesis and high-accuracy transcription using Deepgram for a human-like conversational experience.
• Verified Citations: Every claim is backed by atomic citations linked directly to your source documents to eliminate hallucinations.
• Modern Full-Stack Architecture: A high-performance FastAPI backend paired with a premium Next.js 15+ frontend.

Tech Stack

Backend (Python)

• Framework: FastAPI (Asynchronous, High Performance)
• Package Manager: uv (Extremely fast Python package installer)
• Database: PostgreSQL with pgvector (Vector similarity search)
• Voice Framework: Pipecat & Daily
• AI/LLM: Google Gemini, Voyage AI (Embeddings), Deepgram (STT/TTS)

Frontend (TypeScript)

• Framework: Next.js 15 (App Router)
• Styling: Tailwind CSS with a custom "Void" Dark Theme
• UI Components: Shadcn/ui & Framer Motion
• Auth: Supabase (OAuth & Email)

Getting Started

Prerequisites

• Python 3.11+
• Node.js 18+ & pnpm
• Supabase Account
• API Keys: Gemini, Voyage AI, Deepgram, Daily, Supabase

Installation

1. Clone the repository:

   git clone https://github.com/atharva-again/samvaad.git
   cd samvaad

2. Setup Backend:

   # Install uv if you haven't
   curl -LsSf https://astral.sh/uv/install.sh | sh
   
   # Sync dependencies and create venv
   uv sync

3. Setup Frontend:

   cd frontend
   pnpm install

Configuration

Create a .env file in the root directory (for backend) and frontend/.env.local (for frontend).

Backend .env:

# AI & Pipeline
GEMINI_API_KEY=...
VOYAGE_API_KEY=...
DEEPGRAM_API_KEY=...
DAILY_API_KEY=...

# Database
DATABASE_URL=...
SUPABASE_URL=...
SUPABASE_SERVICE_ROLE_KEY=...

Development

Run Backend

# From the root
uvicorn samvaad.api.main:app --reload

Run Frontend

# From the frontend directory
pnpm dev

Interactive CLI (Internal testing)

# From the root
samvaad

Project Structure

.
├── samvaad/           # Core Backend Logic
│   ├── api/           # FastAPI Routes & App Initialization
│   ├── core/          # Shared Logic (Auth, Memory, Unified Context)
│   ├── db/            # Database Models & Service Layer (pgvector)
│   ├── pipeline/      # RAG Pipeline (Ingestion, Retrieval, Deletion)
│   ├── prompts/       # AI Personas & System Prompt Engineering
│   └── utils/         # Citation Handling, Text Processing, Hashing
├── frontend/          # Next.js Application
├── tests/             # Pytest Unit & Integration Suites
├── PLANS/             # Living Architecture & Development Roadmap
└── AGENTS.md          # Developer Knowledge Base for AI Assistants

Developer Knowledge Base

This project maintains a hierarchical set of AGENTS.md files that act as a "source of truth" for code conventions, architecture, and anti-patterns.

• Root Context: AGENTS.md
• API Context: samvaad/api/AGENTS.md
• DB Context: samvaad/db/AGENTS.md
• Frontend Context: frontend/app/AGENTS.md

License

This project is licensed under the MIT License - see the LICENSE file for details.