Microsoft’s generative-AI platform on Azure gives enterprises access to OpenAI models and many others with Azure’s security, compliance and regional controls. For organisations already running on Microsoft 365 and Azure, it is often the most natural foundation for copilots, RAG applications and agents.
Azure OpenAI Service
Azure OpenAI Service provides access to OpenAI’s models — for chat, reasoning, embeddings, image and speech tasks — hosted on Azure. You deploy models into your own Azure subscription, choose regions and deployment types to meet data-residency and capacity needs, and control access with Entra ID and private endpoints. Microsoft states that customer prompts and completions are not used to train the underlying models.
Azure AI Foundry
Azure AI Foundry is Microsoft’s unified platform for building AI applications and agents. It brings together a model catalogue (OpenAI models plus models from other providers and open-source families), prompt and evaluation tooling, content safety, tracing and an agent service. In practice, Azure OpenAI is the model access layer and Foundry is the workbench around it.
Grounding models with Azure AI Search
Most enterprise solutions combine a model with Azure AI Search for retrieval-augmented generation. AI Search indexes content from SharePoint, Blob Storage, databases and more; supports keyword, vector and hybrid search with semantic ranking; and can apply security trimming so users only retrieve documents they are allowed to see. Combined with Azure AI Document Intelligence for complex PDFs and forms, it covers most knowledge and document use cases.
Security and governance on Azure
- Identity: Entra ID authentication and role-based access instead of shared API keys.
- Network: private endpoints and disabled public access for AI resources.
- Content safety: Azure AI Content Safety filters and prompt-attack protection.
- Data governance: Microsoft Purview for sensitivity labels and data-loss prevention.
- Policy and monitoring: Azure Policy, Defender for Cloud and Azure Monitor.
Copilot Studio or Azure AI Foundry?
Microsoft offers two main ways to build copilots and agents. Copilot Studio is low-code, fast to deliver and well suited to agents that live in Microsoft 365 and Teams. Azure AI Foundry is pro-code, giving engineering teams full control over models, retrieval, orchestration and evaluation for custom applications. Many organisations use both: Copilot Studio for business-led agents, Foundry for complex or customer-facing solutions.
Capacity and cost
Azure OpenAI offers pay-as-you-go pricing and provisioned throughput for predictable, high-volume workloads. Plan quotas early, choose deployment types to match latency and residency needs, and track token usage per application. Prompt efficiency, retrieval quality and model choice are the biggest cost levers.
Our Azure AI team builds secure, governed generative-AI solutions on Azure — from Teams copilots to document automation.
Frequently asked questions
What is the difference between Azure OpenAI and Azure AI Foundry?
Azure OpenAI Service provides access to OpenAI models on Azure. Azure AI Foundry is Microsoft's broader platform for building AI apps and agents, including a model catalogue, evaluation, safety and agent tooling, and it works with Azure OpenAI models.
Does Azure OpenAI use my data to train models?
Microsoft states that prompts and completions sent to Azure OpenAI are not used to train the underlying OpenAI models.
Should I use Copilot Studio or Azure AI Foundry?
Copilot Studio suits low-code agents inside Microsoft 365 and Teams; Azure AI Foundry suits engineering teams building custom, complex or customer-facing AI applications. Many organisations use both.