Before you can call any AI service you need an Azure resource that gives you an endpoint, credentials and a bill. The exam expects you to know the kinds of resources and when each is the right choice. Azure has renamed this area several times: Azure AI Studio became Azure AI Foundry and later Microsoft Foundry, and the resource once called Cognitive Services is now Azure AI services. The resource type in templates is still Microsoft.CognitiveServices/accounts, so you will see the old name in code.
A single-service resource provides one service, such as Language, Speech or Custom Vision, with its own endpoint and keys. It is useful when you want a Free (F0) tier for that one service, separate billing per service, or keys that can only call one API. A multi-service Azure AI services resource exposes many services (Vision, Language, Speech, Translator, Document Intelligence and more) behind one endpoint and key pair with one bill. It is the simplest choice when one app uses several services, and it is the resource you attach to an Azure AI Search skillset to pay for AI enrichment.
Azure AI Foundry is the portal and platform for building generative AI apps and agents. A Foundry project is the workspace where you deploy models, build agents, run evaluations and use playgrounds. Projects can live on an Azure AI Foundry resource (an Azure AI services resource that also supports projects), which is the simpler, newer setup. The older hub-based setup uses an Azure AI hub, which is built on Azure Machine Learning: the hub holds shared settings such as connections to other resources, networking, compute and security, and teams create projects under it. A hub also brings dependent resources such as a storage account and a Key Vault.
Choose based on sharing and isolation. If several teams need common connections and governance but separate workspaces, create one hub (or one Foundry resource) with a project per team or app. If an app only needs to call one or two prebuilt APIs, a single-service or multi-service resource without Foundry is enough. Connections in a project store how to reach other resources, such as an Azure AI Search service or a storage account, so developers do not paste keys into code.
Whatever you create, you get an endpoint URL and, unless key access is disabled, two keys. Resources are created in a region, and not every service or model is available in every region, which is the next topic.
Key terms
- Single-service resource
- An Azure resource for one AI service, with its own endpoint, keys, pricing tier and bill.
- Multi-service resource
- An Azure AI services resource that exposes many AI services behind one endpoint and key pair with combined billing.
- Hub
- A shared Azure AI Foundry container, built on Azure Machine Learning, that holds connections, security and compute for several projects.
- Project
- A Foundry workspace where you deploy models, build agents, evaluate and manage the assets for one app or team.
A consultancy runs three client pilots. It creates one hub with private networking and a connection to a shared Azure AI Search service, then a project per client so each team's deployments, prompts and evaluations stay separate while governance is set once.
Check yourself
An app uses Vision, Language and Translator, and finance wants one bill. What should you create?
A multi-service Azure AI services resource, which gives one endpoint, one key pair and combined billing.
What does a hub provide that a project alone does not?
Shared settings for many projects: connections to other resources, networking, compute and security policies, plus the dependent storage and Key Vault.