AI-103 decision sheets

• Azure AI Content Safety: Prompt Shields vs blocklist vs Task Adherence vs custom categories vs multimodal
• Foundry built-in evaluator categories: quality vs RAG vs safety vs agent
• Foundry agent-building capabilities: memory vs tool catalog vs Foundry IQ vs publishing
• Foundry agent types: prompt vs hosted vs connected
• Foundry RBAC roles: User vs Owner vs Account Owner vs Project Manager vs Agent Consumer
• Which Foundry SDK or client: project client vs OpenAI SDK vs Agent Framework vs Anthropic SDK
• Foundry generative media: text-to-video vs image-to-video vs streaming image
• Reducing fabricated answers: few-shot vs grounding vs fallback vs affordance vs deterministic filter
• Azure Language capability selection: key phrase vs NER vs PII vs sentiment
• Foundry model category: small language model vs multimodal vs frontier vs embedding vs reasoning
• Foundry model deployment: managed compute vs serverless vs instant access
• Azure Speech capability selection: custom speech vs LLM speech vs pronunciation vs language ID vs custom voice
• Image analysis selection: visual question answering vs field extraction vs dense captioning