Multi-Agent Onboarding App (Azure AI Foundry Demo)

This is a demo project showcasing how to use Azure AI Foundry Agent Service to automate an employee onboarding flow using multiple AI agents.

📺 Demo YouTube Video

There’s a quick demo showing the app and how everything works together:
Watch it on YouTube

🧠 What It Does

The app collects an employee's full name, email, and department from a simple frontend form. Behind the scenes, it dynamically creates and coordinates three agents:

• department_rag_agent – Uses Azure AI Search to retrieve relevant onboarding information about the specified department.
• discord_agent – Sends a welcome message to a Discord channel using the employee’s details and department info.
• onboarding_agent – Orchestrates the flow: first runs the RAG search, then delegates to the Discord agent to send the message.

Agents are created, executed, and cleaned up on each request to keep everything stateless.

---

🧾 Required Azure Services

To run this app, you’ll need the following Azure resources:

• Azure AI Foundry Project
  Used to create and coordinate the agents. You’ll use a deployed LLM (e.g., gpt-4o-mini) for agent reasoning and message generation.

• Azure OpenAI
  Used to create and deploy the embedding model (e.g., text-embedding-3-large) for populating the vector index in Azure AI Search.

• Azure AI Search
  Hosts the index with department-specific onboarding content.
  👉 Important: You must connect your AI Search instance to your AI Foundry project.
  Learn how to connect it

• Azure Function App
  Exposes an HTTP endpoint to send Discord messages via webhook. This is necessary because connected agents currently cannot call local Python functions directly.

---

📁 Project Structure

• src/api/ – FastAPI backend with agent definitions, tools, and utilities
• func-api/ – Azure Function App used to send messages to Discord
• web/ – React.js frontend for submitting onboarding info
• data/ – Contains department RAG source files (see below)

---

📄 Data for RAG

The file software_engineering_dept.txt located in the /data folder contains onboarding content for the Software Engineering department. This file is used to populate the Azure AI Search index and powers the department RAG query.

Make sure to:

• Embed the content using the text-embedding-3-small model
• Upload the embedded data into the index defined in your .env file

---

⚙️ .env Configuration

Create a .env file in src/api/ with the following:

# Azure AI Foundry
PROJECT_ENDPOINT=https://<your-foundry-endpoint>/api/projects/<project-name>
MODEL_DEPLOYMENT_NAME=<your-model-deployment-name>

# Azure AI Search
AZURE_SEARCH_ENDPOINT=https://<your-search-endpoint>
AZURE_SEARCH_API_KEY=<your-api-key>
AZURE_SEARCH_INDEX_NAME=<your-index-name>