Quick Revision Overview
This post is a rapid-review summary of AWS managed AI services for exam day. Each service's purpose and key characteristics are organized for fast reference.
---
AWS Managed AI Services at a Glance
Vision (Image & Video)
| Service | Core Function | Keywords | |---|---|---| | Amazon Rekognition | Image/video analysis (faces, objects, text) | Images, faces, video | | Amazon Lookout for Vision | Manufacturing defect detection | Defects, quality |
Text & Language
| Service | Core Function | Keywords | |---|---|---| | Amazon Comprehend | Text analysis (sentiment, entities, language) | Sentiment, entities | | Amazon Comprehend Medical | Clinical info from medical text | Medical, clinical | | Amazon Translate | Automatic translation (75+ languages) | Translation | | Amazon Textract | Extract text, tables, data from documents | Documents, OCR |
Speech
| Service | Core Function | Keywords | |---|---|---| | Amazon Transcribe | Speech to text (STT) | Speech recognition | | Amazon Polly | Text to speech (TTS) | Voice synthesis |
Conversational AI
| Service | Core Function | Keywords | |---|---|---| | Amazon Lex | Chatbot and voice bot builder | Chatbot, dialogue | | Amazon Kendra | Intelligent enterprise search | Internal search |
Recommendations & Forecasting
| Service | Core Function | Keywords | |---|---|---| | Amazon Personalize | Personalized recommendations | Recommendation engine | | Amazon Forecast | Time-series demand forecasting | Demand, inventory |
Anomaly Detection & Fraud
| Service | Core Function | Keywords | |---|---|---| | Amazon Fraud Detector | Online fraud detection | Fraud, transactions | | Amazon Lookout for Metrics | Business metric anomaly detection | KPI anomalies | | Amazon Lookout for Equipment | Industrial equipment anomaly detection | Factory, equipment |
Generative AI & LLMs
| Service | Core Function | Keywords | |---|---|---| | Amazon Bedrock | Single API for multiple foundation models | LLM, foundation model | | Amazon Q | Enterprise generative AI assistant | Enterprise chatbot | | Amazon Titan | AWS-native foundation model | AWS-built model |
Custom ML
| Service | Core Function | Keywords | |---|---|---| | Amazon SageMaker | Full ML lifecycle management | Custom model, pipeline | | SageMaker Canvas | No-code ML model building | No-code ML |
!AWS managed AI services grouped into 8 categories
Easily Confused Services
Rekognition vs Textract: Rekognition analyzes scenes/faces in images; Textract extracts text/data from document images Comprehend vs Kendra: Comprehend analyzes text sentiment/entities; Kendra searches enterprise document repositories Forecast vs Personalize: Forecast predicts numeric time-series values; Personalize recommends items to individual users Transcribe vs Lex: Transcribe converts speech to text; Lex understands conversational intent
---
Exam Key Takeaways
Fully managed services (use without building ML): Image analysis → Rekognition Document extraction → Textract Text analysis → Comprehend Speech to text → Transcribe / Text to speech → Polly Translation → Translate Chatbots → Lex Enterprise search → Kendra Recommendations → Personalize Demand forecasting → Forecast Fraud detection → Fraud Detector
Generative AI: Multiple LLM access → Bedrock Enterprise AI assistant → Amazon Q
Custom ML pipeline → SageMaker