Your company's architecture diagram still says "Azure OpenAI" and "Azure AI Studio". What is Microsoft Foundry, and how does it relate to them?

Microsoft Foundry is Azure's platform for AI applications and agents, and the current name of what was Azure AI Studio and then Azure AI Foundry. Azure OpenAI is now one part of it: OpenAI's models deployed and called through a Foundry resource, beside models from Anthropic, Meta, Microsoft and others. The rename came with real changes: one Foundry resource with projects replaces the old hub plus separate Azure OpenAI and AI Services resources, the Responses API replaces the Assistants API, and a v1 API replaces dated api-version values.

1. Update the diagram

2. What a Foundry project holds

• Model deployments from a catalogue of more than 10,000 models, each with a deployment type (Global Standard, Data Zone, Provisioned, Batch).
• Agents: declarative prompt agents, voice agents, or hosted agents running your own code (Microsoft Agent Framework, LangGraph, Semantic Kernel).
• Tools, retrieval and memory for those agents.
• Governance set once for all of it: Entra ID, role-based access, content filters, private networking, Azure Policy.

3. What it means for existing code

from openai import OpenAI

client = OpenAI(
    base_url="https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/",
    api_key="...",                     # or an Entra ID token provider
)
client.responses.create(model="my-deployment", input="Hello")

An existing Azure OpenAI resource can be upgraded to Foundry keeping its endpoint and keys, so running code keeps working. The real migration work is moving Assistants-based agents to the Responses API.

4. Why it matters

A security team approves one resource and its rules; every model and agent inside inherits them. That is the case for using Foundry rather than separate services per team. See Microsoft Foundry and Azure OpenAI.

What interviewers want: the chain of names, Azure OpenAI as part of Foundry, the single resource with projects, the Responses and v1 API changes, the agent types, governance as the reason it exists, and that upgrading keeps endpoints and keys.