Part 4/11:
Solution Architecture: An AI-Powered Chatbot for Data Queries
The heart of this innovation is an AI-hosted, dynamically deployed chatbot integrated across various dashboards and reporting tools — whether built in React, Angular, PowerBI, or other platforms. Here's an overview of how the system operates:
Data Injection and Embedding
Creation of QA Pairs: The team recorded typical questions users ask and crafted corresponding SQL queries, storing these in a vector database.
Embedding Generation: Questions (not SQL queries) are transformed into vector embeddings, which capture their semantic meaning.
Metadata Enrichment: QA pairs are enhanced with metadata, such as the solution area (procurement, supply chain, etc.) to provide context.