[ ABORT TO HUD ]
SEQ. 1
SEQ. 2
SEQ. 3
SEQ. 4
Platform Architecture Overview
The Foundry Architecture
Microsoft Foundry is a unified Platform-as-a-Service (PaaS) that brings together models, tools, data, agents, and governance under a single Azure resource provider (Microsoft.CognitiveServices).
Core Platform Layers
| Layer | Purpose | Examples |
|---|---|---|
| Models | AI model catalog and deployment | GPT-4o, Llama 3, Mistral, Phi, Cohere |
| Tools | Pre-built AI capabilities (formerly Cognitive Services) | Vision, Speech, Document Intelligence, Translator |
| Data & Grounding | Connect models to your data | Azure AI Search indexes, file uploads, databases |
| Agents | Build and manage autonomous AI agents | Agent Service, Connected Agents, Multi-Agent Workflows |
| Evaluation | Measure quality, safety, and groundedness | Built-in evaluators, adversarial simulation |
| Governance | Security, compliance, and monitoring | RBAC, content filters, tracing, Azure Policy |
Resource Hierarchy
Azure Subscription
└── Resource Group
└── Foundry Resource (Hub)
├── Project A (team workspace)
│ ├── Model Deployments
│ ├── Agents
│ ├── Search Indexes
│ └── Evaluations
└── Project B (another team)
└── ...
💡 Key Insight: The Hub is the organizational container that centralizes governance (RBAC, networking, policies). Projects are isolated workspaces where teams actually build. Projects inherit security settings from their parent Hub.
FOUNDRY VERIFICATION
QUERY 1 // 2
Under which Azure resource provider does Microsoft Foundry operate?
Microsoft.MachineLearning
Microsoft.CognitiveServices
Microsoft.OpenAI
Microsoft.AIFoundry