Domain 1 of 2 · Chapter 3 of 3

Identify AI workloads

Unlock the complete study guide + 1,040 practice questions across 16 full exams.

Bundled into the existing Microsoft Azure AI Fundamentals premium course — no separate purchase.

14-day money-back guarantee — no questions asked.

Included in this chapter:

  • Reading a scenario for its workload
  • Generative AI creates content that was not in the input
  • Agentic AI reasons and acts through tools
  • Text analysis techniques as Azure exposes them
  • Speech: converting between audio and text
  • Computer vision reads images you already have
  • Information extraction with Content Understanding
  • Telling near-miss workloads apart

Which workload family a scenario belongs to

Workload familyPrimary inputRequired outputScenario fingerprintTypical Azure surface
Generative AIA prompt, which can include text, images, or audioNew content that was not present in the input"draft", "compose", "generate an image"A model deployed from the Foundry model catalog
Agentic AIA goal, plus the tools the agent may useA completed task, including actions taken in other systems"look up the order, then update the record"Foundry Agent Service (model, instructions, tools)
Text analysisWritten textLabels, scores, phrases, or a condensed version of that text"how do reviewers feel", "list the main topics"Azure Language in Foundry Tools
SpeechSpoken audio, or text that must be spokenA transcript, a translated output, or synthesized audio"transcribe the calls", "add captions", "read it aloud"Azure Speech in Foundry Tools
Computer visionAn existing imageA description or a location of what the image already contains"identify", "caption", "read the text on the sign"Azure Vision in Foundry Tools, or a multimodal model
Information extractionDocuments, images, audio, or videoValues that already exist in the source, in a defined output shape"populate these fields from every incoming file"Azure Content Understanding in Foundry Tools

Decision tree

Must the output be contentthat was not in the input?YesMust it also take a stepin another system?YesNoAgentic AItask completedGenerative AInew contentNoMust existing values land ina user-defined output schema?YesInformation extractionvalues into a schemaNoIs the required conversionbetween spoken audio and text?YesSpeechaudio to and from textNoDoes the answer depend on animage the caller supplies?YesComputer visionwhat an image showsNoText analysislabels and scores
Cheat sheet

Cheat sheet

  • Generative AI creates new content in response to input
  • Agentic AI reasons and acts across multiple steps by using tools
  • Text analysis identifies language features and meaning in written content
  • Speech workloads convert between spoken audio and text
  • Computer vision interprets objects, text, and visual features in images
  • Information extraction recovers existing source values for downstream use
  • Key phrase extraction returns the main concepts in unstructured text
  • Named entity recognition identifies and categorizes mentions in text
  • Sentiment analysis labels the polarity expressed in text
  • Summarization condenses source content while preserving its main information
  • Keyword extraction corresponds to Azure key phrase extraction
  • Entity detection corresponds to Azure named entity recognition
  • Phrase lists bias speech recognition toward expected terms
  • Batch transcription fits large volumes of prerecorded audio
  • Diarization separates speakers within a transcription
  • Speech translation converts live spoken language into translated text or speech
  • SSML controls how synthesized speech sounds
  • Image classification assigns a category to an image
  • Object detection identifies items and their locations in an image
  • OCR extracts printed or handwritten text from visual content
  • Image analysis returns descriptive features from an existing image
  • Azure Content Understanding processes multiple modalities into a user-defined format
  • A Content Understanding analyzer defines repeatable processing and output fields
  • Content extraction normalizes each modality before field extraction
  • Content Understanding fields can extract, classify, or generate values
  • Confidence scores and source grounding support verification of extracted fields

Unlock with Premium — includes all practice exams and the complete study guide.

Also tested in

References

  1. What is Azure Content Understanding in Foundry Tools?
  2. Study guide for Exam AI-901: Microsoft Azure AI Fundamentals
  3. Introduction to generative AI and agents (Microsoft Learn training module)
  4. What is Azure Language in Foundry Tools?
  5. What is Microsoft Foundry Agent Service?
  6. What is key phrase extraction in Azure Language in Foundry Tools?
  7. What is the named entity recognition (NER) feature in Azure Language?
  8. What is sentiment analysis and opinion mining in Azure Language?
  9. https://learn.microsoft.com/en-us/azure/ai-services/language-service/summarization/overview
  10. Speech to text overview (Azure Speech in Foundry Tools)
  11. Speech translation overview (Azure Speech in Foundry Tools)
  12. Text to speech overview (Azure Speech in Foundry Tools)
  13. What is Image Analysis? (Azure Vision in Foundry Tools)
  14. OCR - Optical Character Recognition (Azure Vision in Foundry Tools)
  15. Migrate from Azure Vision in Foundry Tools - Image Analysis
  16. Azure Content Understanding - What is an analyzer? Configuration and reference