Kamana AI
An AI agent designed for customer support that automates ticket creation and handles banking actions like balance checks and KYC verification through vector search and company data scraping.
Problem
Bank customer support teams face high volumes of routine grievances, leading to delayed response times and manual ticketing overhead.
Solution
Kamana AI acts as a sophisticated frontline agent. It uses Selenium for automatic company data scraping and Milvus for vectorizing the knowledge base. Using a local Gemma 4 26B model via Llama.cpp alongside LangGraph for agentic workflows, the system accurately handles user queries, automatically executes banking functions (account balance, KYC details, blocking), and intelligently generates support tickets when human intervention is required.