CustomerAgent — AI Assistant with RAG

CustomerAgent is a smart virtual assistant that uses Retrieval Augmented Generation (RAG) to answer customer questions based on the company's corporate documentation. It supports multiple AI providers: OpenAI GPT-4o-mini, Groq Llama, Gemini and Claude. The system ingests documents (PDFs, TXT, MD, CSV), processes them into chunks, generates vector embeddings and stores them in Supabase with pgvector. When a customer asks a question, the system retrieves the most relevant chunks and generates a contextual answer using the selected model. It includes an embeddable widget installable on any website, real-time answer streaming, multi-tenant architecture with roles and configurable limits, a plan system with subscriptions, metrics dashboard, conversation history and a documented REST API.

The Challenge

Companies spend hours answering the same customer questions. Traditional chatbots only answer predefined questions and don't understand context.

The Solution

We built a RAG system that ingests corporate documentation, turns it into vector embeddings and generates contextual answers using AI, drastically reducing support time.