Azure AI Foundry Tutorials - Foundry vs Hub Based Projects
Azure AI Foundry is Evolving
In previous tutorial we had created agents in hub based project. Azure AI Foundry is consistently evolving to provide better developer experience and simplified architecture. Microsoft has introduced a new project type called Foundry Projects alongside the existing Hub-based Projects.Understanding the difference between these two project types is important because it impacts how we build AI agents, which endpoints we use, and what features are available to us.
Video
This tutorial is explained in the below Youtube Video.Azure AI Agents - Table of Contents
Azure AI Foundry Hello World Example Azure AI Foundry - Azure AI Agent Hello World Example Azure AI Foundry - Foundry vs Hub Based Projects Azure AI Foundry - Build Agent with Azure AI Foundry SDK Azure AI Foundry - Bing Web Search Agent
Two Types of Projects in Azure AI Foundry
Microsoft Foundry now supports two distinct project types. Let us understand what each type means and how they differ.
Hub-Based Projects (Traditional Approach)
This is what we have been using in previous tutorials. When you create an AI Hub, an Azure AI Foundry resource is automatically provisioned. The architecture looks like this: Hub -> Projects
Hub-based projects have the following characteristics:
- Multiple Azure Resources - When you create a hub, Azure provisions multiple resources including Hub resource, Storage account, Key Vault, Application Insights, and Container Registry.
- Based on Azure Machine Learning - Hub is an implementation of Azure Machine Learning workspace and requires multiple Azure services as dependencies.
- Ideal For - Custom model training, Prompt Flow workflows, fine-tuning models, scenarios requiring maximum control over infrastructure.
Foundry Projects (New Simplified Approach)
This is the newer project type that Microsoft is recommending for most AI agent scenarios. The architecture is simpler: Foundry Resource -> Projects (no hub needed!)
Foundry projects have these characteristics:
- Fewer Resources - Much simpler resource creation. The Foundry resource is created automatically when you create the project. No need to manually provision storage, key vault etc.
- General Availability for Agents - The Azure AI Agent Service is in General Availability (GA) for Foundry projects, while it remains in preview for hub-based projects.
- Future Agent Features - According to Microsoft documentation, all new agents and model-centric capabilities will only land on Foundry projects. This includes access to Foundry Agent Service in GA and Foundry API.
- Ideal For - Building AI agents, working with deployed models, modern AI applications, scenarios where you want simpler setup and management.
Key Differences Side by Side
| Aspect | Hub-Based Project | Foundry Project |
|---|---|---|
| Parent Resource | Azure AI Hub | Azure AI Foundry Resource |
| Resources Created | Many (Hub, Storage, Key Vault, App Insights, Container Registry) | Fewer resources (Foundry resource auto-created) |
| Endpoint Domain | .api.azureml.ms |
.services.ai.azure.com |
| Endpoint Format | https://region.api.azureml.ms/agents/v1.0/subscriptions/.../workspaces/name |
https://resource.services.ai.azure.com/api/projects/name |
| Agent Service Status | Preview | General Availability (GA) |
| Future Agent Features | No (will not receive new agent capabilities) | Yes (all new features will land here) |
| Complexity | Higher (more resources to manage) | Lower (simplified setup) |
| Best For | Custom model training, Prompt Flow, fine-tuning, maximum infrastructure control | AI agents, model deployments, modern AI apps |
| Open Source Models | Supported | Limited support |
| Prompt Flow | Fully supported | Not available |
Microsoft's Official Recommendation
According to Microsoft's official documentation:
"In general, you should use a Foundry project if you're looking to build agents or work with models. Use a hub-based project when you need features that aren't available in a Foundry project."
More importantly for AI agents:
"New agents and model-centric capabilities are only available on Foundry projects, including access to the Foundry API and Foundry Agent Service in general availability."
This means if you are building production AI agents, Foundry projects are the recommended path forward.
When to Use Hub-Based Projects
Hub-based projects are not being deprecated. They continue to be the right choice when you need:- Custom model training capabilities
- Prompt Flow workflows
- Fine-tuning models with Azure ML
- Open source model deployments
- Maximum control over infrastructure and networking
- Enterprise-grade security features with complete VNET integration
When to Use Foundry Projects
Foundry projects are the better choice when you want to:- Build AI agents with production support (GA)
- Deploy and interact with AI models
- Get started quickly with minimal infrastructure setup
- Leverage the latest agent features as they are released
- Use simpler endpoint URLs
- Reduce the number of Azure resources you need to manage
How to Create a New Foundry Project
Creating a Foundry project is simpler than creating a hub-based project. Let us see how to do it. Step 1: Sign in to Microsoft Foundry athttps://ai.azure.com. Click on "Create an agent"
Step 4: Give your project a name and click "Create"
The azure ai foundry project will be created. We can see
- This is ai foundry project
- Agent is not in preview mode.
In the next tutorial, we will be making use of azure ai foundry sdk code to create agent.