This project is an AI-powered Customer Support System designed to intelligently process customer queries for a bank using Prompt Chaining.
It was built as part of KodeCamp 5X to demonstrate structured prompt engineering — showing how AI can interpret, classify, and respond to user messages step-by-step.
The system simulates how a banking chatbot can understand customer messages by passing them through five reasoning stages:
- Interpret the customer’s intent
- Map the query to possible categories
- Select the most appropriate category
- Extract extra information (like amount, card type, or date)
- Generate a professional short response
- Built using ChainForge to simulate AI reasoning
- Uses OpenRouter API to interact with large language models (LLMs)
- Processes queries in five sequential AI reasoning steps
- Fully interactive command-line interface
- Returns clear, structured outputs at every stage
- Node.js (runtime environment)
- Axios (for HTTP requests)
- dotenv (for environment variable management)
- readline-sync (for terminal-based user input)
- OpenRouter AI API (for language model intelligence)
- ChainForge (for visual prompt chain design and testing)
- Clone or download this repository.
- Install all dependencies:
npm install axios dotenv readline-sync