Microsoft Azure AI offerings landscape has been changing at a rapid pace. This change also leads to changes in the nomenclature which creates confusion among the developer community. One of the questions I have come across is the difference between Azure AI Foundry Project and Azure AI Foundry Hub Project. This article is an effort to provide clarity to this question.
Azure AI Foundry supports two types of projects: An Azure AI Foundry project and a Hub-Based project. To get a better understanding of the difference between Azure AI Foundry Project and Azure AI Hub Project, let us take a step back and get some historical perspective. Lets start with the following diagram.
Fig 1: Relationship between Data Science, Machine Learning and AI
You might have come across this diagram before. AI is built on top of ML models. Hence there is an overlap between the toolset used for managing the lifecycle of AI and ML solutions. The same is true for Azure AI toolset as well.
The current Azure ML service (introduced around late 2018) provides a comprehensive, enterprise-ready platform for the entire ML lifecycle, supporting both visual no-code designers. Within this platform, the concept of a “hub” emerged as a foundational workspace for managing security, data access, and core configurations across multiple related projects.
In Azure AI Foundry, the concept of a hub is similarly employed. A Hub groups one or more projects under shared configurations (networking, identity, storage, policy) so that teams or departments can collaborate while inheriting the same enterprise controls.
There is no confusion until here. The confusion starts from the fact that the type of the project created under an AI Foundry resource is different from the project created under a Hub. This can be seen in the resource view in your Azure Portal. In addition to this the project created under Azure AI Foundry resource is not available to be added to a Hub.
The reason for this difference is the historical perspective described earlier. Hub is more of a legacy support carried from Azure machine learning studio. There is no clear strategy from Microsoft on how Hub as a feature will shape up in future foundry releases.
As of now, Azure AI Foundry Project is the recommended option by Microsoft for most of your AI solutions.
Let us understand the details of each project type to understand the difference in more detail.
Azure AI Foundry project
- Azure AI Foundry project is a container for access management, data upload, integration, testing, deployment, and monitoring. This lets you keep your work separated between use cases without needing to create extra Azure resources. Projects function as folders to group related work.
- Azure AI Foundry project is managed under an Azure AI Foundry resource. (Please refer to the article “Azure AI Foundry Overview” for more details in Azure AI Foundry Resource)
- You can create multiple AI Foundry projects on an existing AI Foundry resource to enable team collaboration and shared resource access including security, deployments, and connected tools.
Fig 2: Azure AI Foundry Project Organization
Azure AI Foundry Hub project
- An Azure AI Foundry Hub project is managed under an Azure AI Hub resource. The hub defines the overall boundary for that project’s activities.
- The “Hub Project” was necessary as a bridge to existing or specialized Azure Machine Learning capabilities while Microsoft developed the more integrated, application-centric “Foundry Project” as its go-to platform for general generative AI development
- Hub projects can be used in Azure AI Foundry and Azure Machine Learning studio.
- Azure AI Hub is required for selected use cases. Some of these use cases are Model Fine Tuning, Model Evaluation and Prompt Flow
- Hubs and Projects are containers for project organization. The AI Hub project still requires a connection to the Azure AI Foundry resource to access core AI capabilities and deployed models.
Fig 3: Azure AI Foundry Hub Project Organization
Which type of project do I need?
In general, you should use an Azure AI Foundry project if you are looking to build agents or work with models.
Use a hub-based project when you need features that are not available in an Azure AI Foundry project. See the following table for more on feature availability.
Fig 4: Which project to use when
Azure AI environment is changing at a fast pace and hence it is advisable to check the latest documentation for the most up-to-date offerings.
dabidigital 