LLM Agent on Microsoft Fabric

This code generates a visual or table based on the user input The backend database is Microsoft Fabric SQL warehouse SQL Agent connects to Fabric based on the following contents from .env file

Initialize and Run the Application

Use Docker

docker build -t genaiapp .

docker run -d -p 8501:8501 -e SQL_SERVER_NAME=${sqlservername} \
    -e SQL_SERVER_DATABASE=${sqlserverdatabase} \
    -e GPT4_DEPLOYMENT_NAME=${gpt4deploymentname} \
    -e AZURE_OPENAI_API_VERSION=${azureopenaiapiversion} \
    -e AZURE_OPENAI_ENDPOINT=${azureopenaiendpoint} \
    -e AZURE_OPENAI_API_KEY=${azureopenaiapikey} \
    -e ARM_SUBSCRIPTION_ID=${armsubscriptionid} \
    -e AZURE_TENANT_ID=${azuretenantid} \
    -e AZURE_CLIENT_ID=${azureclientid} \
    -e AZURE_CLIENT_SECRET=${azureclientsecret} \
    -t genaiapp:latest

Run locally

Create an env file following the sample template below

sqlservername={Microst Fabric Warehouse GUID}.datawarehouse.fabric.microsoft.com
sqlserverdatabase={Name of the Microsoft Fabric Warehouse}
gpt4deploymentname={LLM Deployment name}
azureopenaiapiversion={Deployment Version}
azureopenaiendpoint={Azure Open AI EndPoint}
azureopenaiapikey={Azure Open AI Key}
armsubscriptionid={Your Subscription ID}
azuretenantid={Your Azure Tenant ID}
azureclientid={Your SP Client ID}
azureclientsecret={Your SP Secret}

Command

streamlit run app.py

Use your own tables. This fictitious sample data will not be provided with the repository. Couple of sample screen shots this app generates. **** These visualizations are generated with Made UP Data ****

Prompt Sample Input 1:

Create a bar chart to render the average annual income by loan type

Output Visualition

Prompt Sample Input 2:

Create a bar chart showing average credit score by Years of experience

Output Visualization