Brava FAQs

Project Overview & Architecture

Brava FAQs is an AI-powered internal support tool designed to assist call center agents in quickly finding accurate responses to customer queries. The platform integrates semantic search and Retrieval-Augmented Generation (RAG) chat, enabling users to interact with a dynamic knowledge base. Real-time updates are ensured via webhook integration with Brava's CMS. I implemented semantic search using Typesense to surface relevant FAQs based on natural language queries, built a RAG-based chat system using Pinecone for vector search, allowing agents to chat with the knowledge base, integrated webhooks with Brava's CMS for real-time document updates, set up background job processing using Bull queues to support large-scale uploads of knowledge documents, and developed a secure, role-based authentication system to manage user access.