Home > AI Solutions > Gen AI > White Papers > HR-Assist: Proof of Concept RAG-Based Matching Assistant > Setup
After collecting and exporting the data from SmartSheet into a CSV file, it is sent to the HR-Assist chatbot's data processing pipeline in the main.py file. The HR-Assist chatbot is fully containerized using Docker. When the system is launched, the main.py script runs automatically. This script handles the entire data ingestion process with LangChain CSVLoader. The CSVLoader is initialized within the script to read the CSV files directly from their storage directory.
Once initialized, CSVLoader reads the data from these files line by line, converting it into a format that the chatbot’s backend can process. This step is crucial for transforming raw CSV data into structured data frames or objects that the AI models can utilize.
To enhance the effectiveness of the HR-Assist chatbot, we defined the system prompt to focus on several key areas and consider factors like skills and preferences between project submissions and rotation members. This ensures that the most suitable matches are prioritized.
The key matching criteria includes: