3.3.2. Azure AI Services Portfolio
💡 First Principle: Azure AI Services are pre-built, developer-ready AI capabilities that can be embedded in custom applications. They're building blocks—vision, speech, language, decision—that developers use to add AI features without training models from scratch.
The Azure AI Services portfolio includes:
| Service | Capability | Example Use Case |
|---|---|---|
| Azure AI Vision | Image and video analysis: object detection, image tagging, OCR (reading text) | Product recognition, visual inspection, reading scanned text |
| Azure AI Search | Indexes content for keyword, semantic, and vector search, returning the most relevant passages at query time | Enterprise search, knowledge mining, retrieval layer for custom RAG/grounding |
| Azure AI Speech | Speech-to-text, text-to-speech | Transcription, voice assistants |
| Azure AI Language | Text analytics: sentiment, key phrases, entity recognition, language understanding | Sentiment analysis, ticket routing, content classification |
| Azure AI Translator | Text and document translation | Multilingual apps and content |
| Azure AI Document Intelligence | Extract structured data (fields, key-value pairs, tables) from documents | Invoice processing, contract terms, form automation |
| Azure OpenAI (in Foundry Models) | OpenAI models with Azure security, compliance, regional deployment options for data residency, and private networking | Enterprise apps that need GPT models under Azure controls |
These services are consumption-based (pay per use) and designed for developers building custom solutions. They're not productivity tools for end users—they're components for building applications. Each resource is created in an Azure region you choose, so processing can generally stay in-region for data residency, under Azure's enterprise security and compliance commitments.
Pre-built or custom? Use a pre-built service when it meets the requirement; move to Foundry when you need a specific model, fine-tuning on proprietary data, or multi-model agent pipelines the pre-built APIs can't provide.
When to recommend Azure AI Services:
| Scenario | Recommendation |
|---|---|
| End users want AI-assisted productivity | Microsoft Copilot (formerly Microsoft 365 Copilot), not AI Services |
| Developers building custom app with vision | Azure AI Vision |
| Custom application needs document processing | Azure AI Document Intelligence |
| Enterprise search implementation | Azure AI Search |
⚠️ Exam Trap: Azure AI Services are for developers building applications, not for business users wanting productivity assistance. If the scenario describes end-user productivity, the answer is Copilot products, not Azure AI Services.
Reflection Question: A company wants to automatically extract data from incoming invoices and enter it into their accounting system. Would you recommend Microsoft Copilot or Azure AI Document Intelligence?
