cadetAI prototype

Find useful & actionable insights in your customer conversations.

Architecture Overview

cadetAI is a customer conversation analytics platform built with a modern full-stack architecture:

Frontend (Next.js 15)

• Main App (/app/): React-based dashboard with chat interface and document management
• Marketing Site (/website/): Landing page and early access signup
• Components: Reusable UI components using Radix UI and Tailwind CSS
• Authentication: Supabase Auth with protected routes

Backend Services

• Supabase Database: PostgreSQL with vector embeddings for document storage
• Edge Functions: Deno-based serverless functions for document processing
• AI Integration: OpenAI API for embeddings and chat functionality

Data Pipeline

1. Document Ingestion: Import conversations from Intercom API
2. Processing: Split documents into chunks and generate embeddings
3. Tagging: Automatically categorize documents with AI-generated tags
4. Storage: Vector embeddings stored in Supabase with full-text search

Key Features

• Chat Interface: AI-powered conversation analysis with RAG
• Document Management: Upload, process, and search customer conversations
• Intercom Integration: Direct import of customer support conversations
• Vector Search: Semantic search across document embeddings
• Real-time Updates: Live document processing status