Azure AI Agent Service Enterprise Demo Changelog

1.1 (2025-02-08)

Features / Enhancements

Optional Direct Azure AI Search Integration
- Thanks to @farzad528 for adding a feature that allows direct Azure AI Search usage alongside the existing vector store approach.
- The notebook logic now checks if a FileSearchTool is present; if not, it configures Azure AI Search using AZURE_SEARCH_CONNECTION_NAME and AZURE_SEARCH_INDEX_NAME from your .env.
Logic App Integration for send_email
- Replaced the local/mocked send_email function with an HTTP call to a Logic App.
- Added a LOGIC_APP_SEND_EMAIL_URL parameter to .env.example, along with instructions and an ARM template (send_email_logic_app.template.json) for deploying the Logic App.
Environment Configuration Updates
- Revised .env.example to unify all optional parameters (Bing, Logic App, Azure Search, etc.) in one place.
- README now describes how to set these environment variables and clarifies how each integration (Bing, Logic App, Azure AI Search) is triggered.

Bug Fixes

Correction of swapped OpenWeather parameters
- Thanks to @gerbermarco for fixing the OPENWEATHER_ONE_API_KEY and OPENWEATHER_GEO_API_KEY variable values in .env.example.

Breaking Changes

Features

Initial release of an enterprise-grade streaming agent built on Azure AI Agent Service.
Demonstrates programmatic creation or reuse of an agent model (e.g., GPT-4o).
Integrates local enterprise data (HR, PTO, policy files) into a vector store for retrieval-augmented generation (RAG).
Offers optional Bing grounding and custom Python functions (e.g. weather, stock lookup, email sending).
Shows how to stream partial responses and tool calls in real-time.
Includes a Gradio interface for interactive demos.

Bug Fixes
None

Breaking Changes
None