Introduction
Azure AI Foundry is a platform for AI development on Microsoft Azure. While you can provision individual Azure AI services resources and build applications that consume them without it, the project organization, resource management, and AI development capabilities of Azure AI Foundry make it the recommended way to build all but the simplest solutions.
Azure AI Foundry provides the Azure AI Foundry portal, a web-based visual interface for working with AI projects. It also offers the Azure AI Foundry SDK, which you can use to build AI solutions programmatically.
Supported Resource Types
Azure AI Foundry Portal and SDK clients support multiple Azure resource types, each tailored for specific development and operational needs:
Azure AI Foundry New !!
The Microsoft Build last month had some exciting updates to the Azure AI Foundry service, including the Azure AI Foundry Resource.
Azure AI Foundry resource is the next version of the former Azure AI Services.
- It retains all previous capabilities such as Azure OpenAI, models for Speech, Vision, and Language.
- New features include:
- Agent Service
- Evaluations
- Projects (organized as folders)
- Connections to integrate with other Azure services
- AI Foundry API to interact with all new capabilities
Microsoft recommends using the AI Foundry resource (
AIServices) whenever possible to access multiple Azure AI services with a single set of credentials. For services not yet available via AI Foundry (e.g., Face and Custom Vision), you can create a single-service resource instead. ^1
Azure AI Hub
Azure AI Hub resource is used alongside Azure AI Foundry to access:
- Open-source model hosting
- Fine-tuning capabilities
- Azure Machine Learning features
- Creating an AI Hub also automatically creates an Azure AI Foundry resource
- Usable in both the AI Foundry Portal and Machine Learning Studio
Azure OpenAI
A specialized resource for accessing OpenAI models like GPT-4 and GPT-4o.
Offers a subset of Azure AI Foundry capabilities focused on OpenAI APIs.
Azure AI Search
A resource used to index and retrieve data for grounding AI applications. It can be connected to Azure AI Foundry agents to enable retrieval-augmented generation (RAG) and semantic search experiences.
Project Types in Azure AI Foundry
Projects are the core units of development in Azure AI Foundry, accessible via the portal or SDK.
Foundry-Based Project
- A Foundry-based project is associated with an Azure AI Foundry resource in an Azure subscription. Foundry projects support Azure AI Foundry models (including OpenAI models), the Azure AI Foundry Agent Service, Azure AI services, and tools for evaluation and responsible AI development.
Hub-Based Project
- Hub-based projects are associated with an Azure AI Hub resource and include an Azure AI Foundry resource, as well as managed compute, support for Prompt Flow development, and connected Azure Storage and Azure Key Vault resources for secure data storage. Azure AI Hub resources support advanced AI development scenarios, such as developing Prompt Flow-based applications or fine-tuning models. These projects can be used in both the Azure AI Foundry portal and the Azure Machine Learning portal.
Capability Comparison: Foundry Project vs Hub-Based Project
| Capability | Foundry Project | Hub-Based Project |
|---|---|---|
| Agents | ✅ (GA) | ✅ (Preview only) |
| AI Foundry API to work with agents and across models | ✅ (Native support) | Available via connections |
| Models sold directly by Azure - Azure OpenAI, DeepSeek, xAI, etc. | ✅ | Available via connections |
| Partner & Community Models sold through Marketplace - Stability, Bria, Cohere, etc. | ✅ | Available via connections |
| Open source models e.g. HuggingFace | ✅ | ✅ |
| Evaluations | ✅ | ✅ |
| Playground | ✅ | ✅ |
| Prompt flow | ❌ | ✅ |
| Content understanding | ❌ | ✅ |
| Project files (directly upload files and start experimenting) | ✅ | ❌ |
| Project-level isolation of files and outputs | ✅ | ✅ |
| Required Azure dependencies | - | Azure Storage account, Azure Key Vault |
Summary
- The new Azure AI Foundry resource greatly simplifies infrastructure setup for AI application development.
- An Azure AI Hub resource and Hub-based project may not be necessary unless your use case involves capabilities not available in a Foundry-based project.