AI

Not for Azure OpenAI. A Microsoft.CognitiveServices/accounts with kind: OpenAI, or an app-code workload whose ai_source is azure_openai/openai/anthropic, routes to the Bedrock guides, not here — see § LLM Routing.

LLM Routing

If the detected AI workload is LLM-based (generative text/chat), load the source-specific Bedrock guide instead of this file:

• summary.ai_source == "azure_openai" or "openai" → vendored/ai/ai-openai-to-bedrock.md (source-cloud-agnostic — its header explicitly serves Azure OpenAI, since the Bedrock target does not depend on which endpoint served the calls).
• summary.ai_source == "anthropic" → vendored/ai/ai-anthropic-to-bedrock.md.
• summary.ai_source == "both" → both of the above.
• summary.ai_source == "other" or absent, OR the workload is traditional ML (Vision, Document Intelligence, Speech, Language, Translator, custom Azure ML models) → use the SageMaker/Rekognition/Textract/Comprehend/Transcribe/Translate/Polly rubric below.

A generative ai_source does not exempt a traditional-AI workload: an azure_openai codebase that also calls Azure AI Document Intelligence needs BOTH the OpenAI guide (for its GPT workload) and this file (for the Document Intelligence workload). Capability is evaluated per workload.

Signals (Decision Criteria)

Azure Machine Learning (Microsoft.MachineLearningServices/workspaces, online/batch endpoints, jobs)

• Custom model inference (online endpoint) → SageMaker Endpoints.
• Pre-built / catalog model APIs → the AWS AI API for that capability (Rekognition, Textract, Comprehend, Transcribe, Translate, Polly).
• Batch scoring / batch endpoint → SageMaker Batch Transform.
• Training jobs, AutoML → SageMaker managed training / SageMaker Autopilot.

Azure AI Vision (Computer Vision) — kind: ComputerVision

• Image classification, object/label detection, moderation, faces → AWS Rekognition.
• OCR / read (printed or handwritten text in images) → AWS Textract (DetectDocumentText) for document OCR; Rekognition DetectText for scene text in images.

Azure AI Document Intelligence / Form Recognizer — kind: FormRecognizer

• Form / invoice / receipt / ID / business-card extraction → AWS Textract (AnalyzeExpense for invoices/receipts, AnalyzeID for identity documents, or AnalyzeDocument with FORMS/TABLES for general structured extraction).
• General document OCR → AWS Textract (DetectDocumentText).
• Custom/trained document models → Textract custom queries, or SageMaker if the model is a genuinely custom vision model rather than a Document-Intelligence template.

Azure AI Speech — kind: SpeechServices

• Audio/video transcription (batch or streaming) → AWS Transcribe (StartTranscriptionJob / StartStreamTranscription).
• Call-centre audio with sentiment/topic needs → AWS Transcribe Call Analytics.
• Text-to-speech / voice synthesis → AWS Polly (a new, separate line item — Azure bundles STT and TTS under one Speech resource; on AWS they are Transcribe and Polly respectively).

Azure AI Language (Text Analytics) — kind: TextAnalytics

• Sentiment, key-phrase, entity/PII recognition, language detection, summarization → AWS Comprehend (or Comprehend Medical for PHI-bearing clinical text).
• Custom text classification / custom NER → Comprehend custom classifiers/entity recognizers, or SageMaker for a fully custom model.

Azure AI Translator — kind: TextTranslation

• Text translation → AWS Translate.

6-Criteria Rubric

Apply in order, first match wins (the rubric selects a SERVICE; sizing is post-selection):

1. Eliminators: does the Azure config require an AWS-unsupported feature? If yes, note the gap and pick the closest alternative (or defer).
2. Operational Model: managed (SageMaker / the managed AWS AI APIs) vs custom (EC2 + training). Prefer managed.
3. User Preference: preferences.json → design_constraints.cost_optimization + ai_constraints (if present). Cost-sensitive → prefer SageMaker Spot + Autopilot, and the pay-per-call AI APIs over always-on endpoints.
4. Feature Parity: does the Azure config need a model type unavailable on AWS? (e.g. a custom framework — most map to SageMaker containers.)
5. Cluster Context: are other resources in the cluster running ML on SageMaker? Prefer SageMaker affinity.
6. Simplicity: managed AI APIs / SageMaker endpoints over custom EC2.

Right-Sizing

Post-selection, structurally parallel to the compute rubric’s ## Right-Sizing. The managed AWS AI APIs (Rekognition, Textract, Comprehend, Transcribe, Translate, Polly) are pay-per-call and carry no instance size — set sizing_provenance accordingly and do not invent a capacity. A SageMaker endpoint DOES carry an instance type: size it from observed traffic when utilization exists (sizing_provenance: measured), else state a dev-tier default (ml.m5.large) with sizing_provenance: model_prior.

CPU Architecture

Not applicable to the managed AI APIs. A SageMaker endpoint follows the skill’s x86_64 default unless a Graviton-supported container image and a stated preference both hold.

Examples

Example 1 — Azure ML online endpoint (custom PyTorch model) → SageMaker Endpoint

azure_type: Microsoft.MachineLearningServices/workspaces/onlineEndpoints, azure_config: { framework: "PyTorch", version: "2.1" } → aws_service: "SageMaker", aws_config: { endpoint_name: "...", instance_type: "ml.m5.large", container_image: "pytorch:2.1" }, confidence: inferred, sizing_provenance: model_prior, rationale: "Azure ML custom online endpoint → SageMaker Endpoint (PyTorch supported)".

Example 2 — Azure AI Vision (image labelling) → Rekognition (capability image_analysis)

azure_type: Microsoft.CognitiveServices/accounts (kind: ComputerVision) → target_aws_service: "rekognition", target_bedrock_model: null, confidence: inferred, honest_assessment: "not_applicable", rationale: "Azure AI Vision label/object detection → Rekognition DetectLabels".

Example 3 — Azure AI Document Intelligence (invoices) → Textract (capability document_extraction)

kind: FormRecognizer, invoice/receipt extraction → target_aws_service: "textract", aws_config: { api: "AnalyzeExpense" }, confidence: inferred, honest_assessment: "not_applicable".

Example 4 — Azure AI Language (sentiment/entities) → Comprehend

kind: TextAnalytics → target_aws_service: "comprehend", confidence: inferred, honest_assessment: "not_applicable", rationale: "Azure AI Language sentiment + entity recognition → Amazon Comprehend". This section has no gcp analogue — it is net-new for Azure.

Example 5 — Azure AI Speech (transcription) → Transcribe (capability speech_transcription)

kind: SpeechServices, batch transcription → target_aws_service: "transcribe", confidence: inferred, honest_assessment: "not_applicable". If the resource also does synthesis, emit a SECOND design_block mapping the TTS capability to Polly.

Output Schema

Each mapped resource contributes a design_blocks[] row (see references/shared/schema-design-aws-ai.md). Traditional-AI workloads set target_aws_service (one of textract, rekognition, comprehend, transcribe, translate, polly, sagemaker), leave target_bedrock_model: null, and carry honest_assessment: "not_applicable" — they are feature/service swaps, not Bedrock model migrations. Confidence is inferred (or measured when utilization backed a SageMaker size). azure_id / azure_type / azure_config carry the source facts.