In this project, I developed an AI-powered system using ChatGPT with function calling to extract and process information based on user interactions. The system, implemented in Python, was designed to handle complex workflows and automate processes efficiently. A key aspect of this project involved fine-tuning the language model to align with specific project needs, ensuring accurate intent recognition and data extraction.

For the backend, the solution was initially hosted on Heroku for testing and development purposes, allowing for agile testing and deployment. Once the testing phase was completed, the production environment was moved to Azure, ensuring scalability and robustness in a live setting.

The chatbot used a custom-built intent recognition pipeline, integrated with various APIs, and designed to interpret user messages, extract relevant information, and automate specific actions based on the detected intent. By utilizing functions like process_intention and integrating a conversational QA retriever, the system could handle both simple and complex queries dynamically, all from a vector database.

The system integrated various APIs to process user requests, custom made vector database and utilizing tools like function calling or later renamed Tools and managing order-related workflows using custom Python scripts.

The code included structured classes to manage conversations, order processing, and data persistence, with a focus on real-time performance and a seamless user experience. Additionally, the solution incorporated cloud-based data management and error logging, ensuring that all interactions were tracked and processed correctly.

By implementing this AI-driven approach, I was able to significantly enhance workflow efficiency and provide an intelligent, automated solution tailored to the project’s unique requirements.