AWS AI services cheat sheet: 9 services explained with everyday life
Rekognition, Textract, Comprehend, Translate, Transcribe, Polly, Lex, Kendra and Personalize, each explained with an everyday analogy and the exam keywords that point to it.
One type of question shows up a lot on the AWS Certified AI Practitioner (AIF-C01): a short business scenario, and you have to pick the right AWS AI service. The services have similar names, so I built this cheat sheet with an everyday "job" for each one.
The good news: these are ready-to-use AI services. You don't need to train a model; you call an API.
👁️ Amazon Rekognition: the security guard who recognizes faces
Analyzes images and videos: detects objects, people, text in images, faces and inappropriate content.
Exam keywords: image moderation, face comparison, detect objects in photos or video.
📄 Amazon Textract: the assistant who types your paper forms
Extracts text, forms and tables from documents, going beyond simple OCR by understanding the structure (keys, values, cells).
Exam keywords: scanned documents, invoices, forms, extract data from PDFs.
💬 Amazon Comprehend: the analyst who reads customer reviews
Natural language processing on text: sentiment, entities, key phrases, language detection and PII detection.
Exam keywords: sentiment analysis, classify customer feedback, extract entities from text.
🌍 Amazon Translate: the interpreter
Translates text between languages.
Exam keywords: translate product descriptions, multilingual support for text.
🎙️ Amazon Transcribe: the court stenographer
Converts speech to text.
Exam keywords: call center recordings, subtitles, transcribe meetings.
🔊 Amazon Polly: the audiobook narrator
The opposite of Transcribe: converts text to lifelike speech.
Exam keywords: read articles aloud, voice for an app, text-to-speech.
Memory trick: Transcribe writes down what you say; Polly says what you wrote.
🤖 Amazon Lex: the receptionist
Builds conversational interfaces (chatbots and voice bots) that understand intent and collect information, the same technology behind Alexa.
Exam keywords: chatbot, voice bot, book appointments through conversation.
📚 Amazon Kendra: the librarian who understands your question
Intelligent enterprise search across company documents, answering natural language questions. It can also be used as a retriever in RAG solutions.
Exam keywords: search internal documents, enterprise search, natural language questions over company data.
🛍️ Amazon Personalize: the bookstore clerk who knows your taste
Creates personalized recommendations based on user behavior, the same kind of technology Amazon.com uses.
Exam keywords: product recommendations, "customers who bought this also bought", personalized content.
Where do Bedrock and SageMaker fit?
- Amazon Bedrock: when the scenario needs generative AI with foundation models (write, summarize, chat, RAG).
- Amazon SageMaker AI: when you need to build, train and deploy your own models.
- The services above: when there's a specific, common task that an existing API already solves.
💡 On the exam: look for the input and output. Image → Rekognition. Document → Textract. Text meaning → Comprehend. Audio → text: Transcribe. Text → audio: Polly. Language A → B: Translate. Conversation → Lex. Search → Kendra. Recommendations → Personalize.
Quick summary
- Rekognition: security guard (images and video)
- Textract: form-typing assistant (documents)
- Comprehend: review analyst (text and sentiment)
- Translate: interpreter (languages)
- Transcribe: stenographer (speech to text)
- Polly: narrator (text to speech)
- Lex: receptionist (chatbots)
- Kendra: librarian (enterprise search)
- Personalize: bookstore clerk (recommendations)
Your turn
Which pair of services do you mix up the most? Mine was Transcribe and Polly. Tell me yours in the comments! 📚
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article