Skip to content

Select the Appropriate Azure AI Foundry Services

This section of the Microsoft AI-102: Designing and Implementing a Microsoft Azure AI Solution exam covers identifying and selecting the right Azure AI Foundry services for various solution domains. Below are study notes for each sub-topic, with links to Microsoft documentation, exam tips, and key facts.

Select the Appropriate Service for a Generative AI Solution

📖 Docs: Architecture Center: Choose an Azure AI services technology

Overview

  • Generative AI solutions require large language models (LLMs) or other generative models (image, text, code).
  • Primary service: Azure OpenAI Service (models like GPT-4, GPT-3.5, DALL·E, Whisper)
  • Managed through Azure AI Foundry hubs/projects

Key Points

  • Use Azure OpenAI for text generation, summarization, translation, code generation, chatbots
  • Model catalog in Azure AI Foundry provides additional generative models (e.g., Phi, Mistral)
  • Responsible AI must be applied (content filters, monitoring)

Exam Tip

If the scenario involves text, chat, or image generation, Azure OpenAI is the service.

Select the Appropriate Service for a Computer Vision Solution

📖 Docs: Architecture Center: Choose an Azure AI services technology

Overview

  • Computer Vision solutions process and analyze images or video
  • Primary service: Azure AI Vision
  • Capabilities:
    • Object detection, classification
    • OCR (Optical Character Recognition)
    • Spatial analysis
    • Face detection/recognition (legacy)

Key Points

  • For OCR → Read API in AI Vision
  • For custom models → Custom Vision (training image classification or object detection models)
  • Video Indexer can analyze video/audio at scale

Limits

Some features (e.g., facial recognition) have restricted access due to Responsible AI concerns.

Select the Appropriate Service for a Natural Language Processing Solution

📖 Docs: Architecture Center: Choose an Azure AI services technology

Overview

  • Natural Language Processing (NLP) = understanding and extracting meaning from text
  • Primary service: Azure AI Language
  • Capabilities:
    • Key phrase extraction
    • Sentiment analysis
    • Named entity recognition
    • Text summarization
    • Language detection

Key Points

  • Use Custom Text Classification for domain-specific models
  • Supports multilingual processing
  • Integrates with AI Foundry workflows

Exam Tip

If the scenario asks for sentiment, classification, or entity recognition, select AI Language.

Select the Appropriate Service for a Speech Solution

📖 Docs: Architecture Center: Choose an Azure AI services technology

Overview

  • Speech solutions convert between spoken audio and text
  • Primary service: Azure AI Speech
  • Capabilities:
    • Speech-to-text (STT)
    • Text-to-speech (TTS)
    • Speech translation
    • Speaker recognition

Key Points

  • Supports custom voice models
  • Real-time and batch transcription supported
  • Can integrate with call centers, IVR, accessibility tools

Use Case

  • Call center transcription
  • Voice-enabled chatbots

Select the Appropriate Service for an Information Extraction Solution

📖 Docs: Architecture Center: Choose an Azure AI services technology

Overview

  • Extracts structured information from documents
  • Primary service: Azure AI Document Intelligence (formerly Form Recognizer)
  • Capabilities:
    • Prebuilt models (invoices, receipts, IDs, business cards)
    • Custom document extraction models
    • Layout extraction

Key Points

  • Best for semi-structured or unstructured documents (PDFs, images)
  • Custom models require training with sample documents
  • Works with both forms and free-text documents

Exam Tip

Look for keywords: invoices, receipts, IDs, forms → Document Intelligence.

Select the Appropriate Service for a Knowledge Mining Solution

📖 Docs: Architecture Center: Choose an Azure AI services technology

Overview

  • Knowledge mining = extracting insights from large document/data collections
  • Primary service: Azure AI Search
  • Works with:
    • Cognitive skills (OCR, entity recognition)
    • Indexes and vector search
    • Integration with RAG for generative AI

Key Points

  • Can enrich data using AI enrichment pipeline
  • Integrates with Azure OpenAI for retrieval-augmented generation (RAG)
  • Ideal for enterprise search portals

Use Case

  • Corporate knowledge base search.
  • Document repository exploration.

Quick‑fire revision sheet

  • 📌 Generative AI → Azure OpenAI
  • 📌 Computer Vision → AI Vision (Custom Vision for custom models)
  • 📌 NLP → AI Language
  • 📌 Speech → AI Speech
  • 📌 Information Extraction → Document Intelligence (Form Recognizer)
  • 📌 Knowledge Mining → AI Search