Plan and manage an Azure AI solution 1.2 Set up AI solutions in Foundry

Three teams will build and evaluate their own agents.

Study guide skill: Design Azure infrastructure for AI apps and agent-based solutions

Question B8-004

Three teams will build and evaluate their own agents. They need isolated files, agents, and evaluations, but should reuse the same model deployments and connections under centralized governance and networking. How should you structure the Foundry environment?

Select one answer.

  • A A standalone Azure OpenAI resource for each team
  • B One Foundry resource with a single project that all three teams share
  • C One Foundry resource with a separate project for each team ✓ Correct answer
  • D A separate Foundry resource for each team, each with its own model deployments

Answer: C

The Foundry resource scopes governance such as security, connectivity, and deployments, while projects isolate development with boundaries for access control, files, agents, and evaluations. Projects reuse existing model deployments and connections, so one resource with a project per team fits multi-team access. A standalone Azure OpenAI resource suits completions without agent hosting or evaluation, and separate resources would duplicate deployments and governance.

Deep dive

How it works

Foundry uses a layered architecture: a top-level Foundry resource for governance, projects for development isolation, and connected Azure services for storage, search and secrets.

  • Foundry resource: scopes management operations such as configuring security, connecting to other Azure services and managing deployments. Networking, security and model deployments are governed here.
  • Project: a development boundary inside the resource. Projects isolate development activities and provide boundaries for access control, files, agents and evaluations. Teams work in a preconfigured environment, reusing existing model deployments and connections without repeated IT setup.

That separation lets IT apply centralized controls at the resource level while teams work within project-level boundaries. Microsoft lists this exact scenario as a reason to use the model: "Multiple teams need isolated projects with shared model deployments and centralized governance."

So: one Foundry resource (shared deployments, connections, networking, governance) and one project per team (isolated files, agents, evaluations).

The other options, and when you'd use them

  • Standalone Azure OpenAI resource per team: Microsoft says a standalone Azure OpenAI resource might be enough only if the workload needs completions without agent hosting or evaluation. These teams build and evaluate agents.
  • One shared project: Projects are the isolation boundary for files, agents and evaluations. One project gives the teams no isolation from each other.
  • Separate Foundry resource per team: Deployments and governance live at the resource level, so separate resources split them up instead of sharing and centralizing them.

Exam tip

Foundry resource = governance, networking, deployments. Project = per-team isolation of files, agents, evaluations, reusing the resource's deployments.

References

Last reviewed:

Ready to test yourself? The quiz asks this and every other AI-103 question, with instant feedback. Practice this in a quiz →

All Plan and manage questions · AI-103 study guide

Thanks for your feedback!