Overview

An RAG based AI agent trained on a Capital Project database. It can answer questions based on the data stored in the tables and can generate relevant visualization if the data qualifies for it.

• Uses Vector embeddings to fetch relevant metadata.
• Supports mySQL and PostGres Database.

As seen in the above snapshots, the agent takes a question, converts the question into a SQL query, runs the query in the designated database, and fetches the results in tabular as well as the appropriate visualization format. The agent is pre-trained on the required tables in the database and is able to understand the column names to run the query as well as generate an automatic visualization. Additionally, it summarises the tabular info as well for easy understanding.