Relayt 🚀

Relayt is a comprehensive AI-powered Customer Support SaaS platform. It is designed to provide businesses with an intelligent chatbot that trains on their custom datasets, captures leads effectively, and offers a real-time dashboard where human agents can monitor and seamlessly take over conversations.

📖 Key Features & What It Solves

• AI Trained on Your Dataset: The AI doesn't just give generic answers; it replies directly based on the dataset (documents, helpdesk Q&As, scraped website data via namespaces) provided by the business.
• Lead Capture & Real-Time Dashboard: The platform captures customer leads the moment they interact with the widget. Business owners can monitor all active user chats in real time from their admin dashboard.
• Seamless Human Handoff (Live Chat): Whenever the AI can't resolve an issue, or if the user wants direct communication, the agent can switch modes to human. From there, human agents can natively chat with customers directly through the dashboard.
• Customizable Website Embedding: We provide an easily embeddable widget that businesses can place right on their websites. The widget is highly customizable—businesses can adjust themes, bot avatars, initial greetings, placeholders, and quick replies to match their specific brand identity.
• Audio & Video Calls: working on it
• Email Campaigns & Notifications: Easily execute marketing campaigns to captured leads and integrate WhatsApp or Resend to get instantly notified on important customer interactions.
• SaaS Billing: Completely automated subscription processing handled through Stripe (switching to razorpay).

🛠️ Tech Stack Used

Relayt is engineered to support a highly concurrent and real-time load natively:

Backend

• Node.js & Express.js: The core backend HTTP layer.
• TypeScript: Ensuring type safety across the entire application ecosystem.
• Prisma (PostgreSQL): Robust relational database mapping handling complex multi-tenant SaaS schema.
• Pinecone, LangChain & LLMs (Groq/Google GenAI): Powering the conversational AI, vector embeddings, and RAG-based dataset answering architectures.
• Stream SDK & Pusher: Powering ultra-low latency real-time chats, live dashboard updates, and A/V calling infrastructure.
• Clerk (@clerk/express): Enterprise-grade authentication and identity management.
• Stripe: Handling SaaS tier plans, subscription states, and credit billing.
• Firecrawl: Automated web scraping for instant knowledge base data generation.

Frontend

• React 19 & Vite: Lightning-fast frontend tooling and rendering.
• TypeScript: Strict type schemas shared directly with the backend.
• Tailwind CSS & Framer Motion: For robust, deeply customizable widget themes and cinematic SaaS dashboard transitions.
• Zustand: Lightweight, unopinionated global state management.
• Radix UI & Lucide Icons: Accessible, headless UI components.

📂 Project Structure

• /Backend: Contains the Express server, Prisma schema, webhook controllers (Stripe/Clerk), AI/Langchain intelligence logic, real-time socket endpoints, and background worker logic.
• /Frontend: Contains both the robust SaaS dashboard interface and the embeddable standalone chat widget logic.

🚀 Getting Started

Prerequisites

• Node.js (v18+)
• Active PostgreSQL Database
• API Keys: Stripe, Clerk, Pinecone, Firecrawl, Stream, and Groq/Google GenAI

Installation

1. Clone the repository

   git clone https://github.com/Viishal-62/Relayt.git
   cd Relayt

2. Backend Setup

   cd Backend
   npm install
   # Configure your .env file with Stripe, Clerk, PostgreSQL, Pinecone, Stream, etc.
   npx prisma generate
   npx prisma db push
   npm run dev

3. Frontend Setup

   cd ../Frontend
   npm install
   # Configure your .env securely
   npm run dev

🤝 Contributing & Adding Features

We welcome contributions! For adding other features:

1. Please do not push directly to the main branch.
2. Create a new dedicated feature branch: git checkout -b feature/your-awesome-feature.
3. Commit your changes and push the branch.
4. Open a Pull Request outlining your architectural changes.