Native Hosting for Lightweight Applications (e.g., Streamlit) Within Microsoft Fabric

Today, when data scientists or analysts want to share interactive prototypes (e.g., Streamlit apps) built from Fabric Lakehouse or Warehouse data, they need to deploy the app externally in Azure services.
This adds friction:

• Managing Azure App Service / Container Apps
• Setting up authentication (AAD configuration, tokens, redirects)
• Handling deployment pipelines
• Managing app infrastructure, monitoring, scaling
• Needing specialized DevOps knowledge

For simple prototypes or internal tools, this overhead slows innovation and discourages rapid experimentation.

Proposed Solution

Introduce a Fabric-native lightweight application host that:

• Allows publishing small Python-based apps (Streamlit/Gradio/Dash) directly inside Fabric
• Automatically handles authentication via Fabric’s built-in identity layer
• Runs apps securely within the Fabric tenant with data access governed by OneLake permissions
• Requires zero DevOps or infrastructure setup
• Provides an endpoint or embedded viewer so apps can be shared like a Power BI report
• Supports scheduled / event-driven refresh for apps that compute results
• enable closed-loop, input-driven analytics for data science and operational decisioning.

This feature would significantly accelerate:

• Data science prototyping
• Interactive analytics apps
• ML model demos and dashboards
• Lightweight internal tools
• Closed loop analytics

It keeps everything inside Fabric—compute, storage, security, identity—without requiring additional Azure resources or specialized app hosting knowledge.

Benefits

• Rapid prototyping with immediate value
• Lower cost (no extra Azure services)
• Consistent governance and security
• Fewer deployment steps
• Better collaboration among data teams

Example Use Cases

• Streamlit app showing ML model predictions from a Lakehouse table
• Internal parameterized calculators built by data analysts
• Exploratory data web apps for business teams
• Scenario builder using user inputs
• Apps that take input → run logic → write results back