Pre-requisites:

• SQL Database (if using sample data, use Azure SQL with Adventure Works dataset)
• Azure AI Search

Instructions

1. Create an Azure SQL DB using the Adventure Works Sample Dataset (https://learn.microsoft.com/en-us/sql/samples/adventureworks-install-configure?view=sql-server-ver16&tabs=ssms)

2. Update .env and constants.py with your environment variables

3. Gather sample questions and answers and put in a CSV (as shown in src/data/sql_query_examples.csv)

4. Extract Schema from SQL DB - may need to be updated depending on customer scenario. The purpose of this step is to extract a schema description into a .txt file to be used as context for LLM calls for SQL generation.

   python extract_descriptions.py

5. Generate vector database

   python generate_ai_search_vectors.py

6. Command to start the application streamlit run app.py

Authentication Options

Networking Guidance / Recommendations

TO DO

• Update authentication
• Add instructions
• auth options - use entra ID rather than DB password
• Consider MySQL, Postgres (could be long-term)
• Networking - currently it is whitelisted