[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"cheat-sheet---en":3,"domain-info---en":3,"topic-info----en":3,"next-aws-certified-cloud-practitioner-cloud-technology-and-services-ai-ml-and-analytics-ai-ml-services-en":4,"prev-aws-certified-cloud-practitioner-cloud-technology-and-services-ai-ml-and-analytics-ai-ml-services-en":18,"lesson-aws-certified-cloud-practitioner-cloud-technology-and-services-ai-ml-and-analytics-ai-ml-services-en":28},null,{"locked":5,"reason":6,"meta":7,"item":3},true,"paywall",{"title":8,"description":9,"isFree":10,"estimatedMinutes":11,"difficulty":12,"learningObjectives":13},"AWS Analytics Services","Learn the core AWS analytics services: Amazon Athena, Redshift, EMR, AWS Glue, Amazon Kinesis and Data Firehose, QuickSight, Lake Formation, OpenSearch Service, and how they fit into a data pipeline.",false,18,"beginner",[14,15,16,17],"Describe the stages of a typical AWS analytics pipeline: ingest, store, process, query, and visualize","Match the core analytics services to the job each one does","Compare Amazon Athena, Amazon Redshift, and Amazon EMR","Identify which services handle real-time streaming data",{"locked":5,"reason":6,"meta":19,"item":3},{"title":20,"description":21,"isFree":10,"estimatedMinutes":22,"difficulty":12,"learningObjectives":23},"Hybrid and Private Connectivity","Learn how to connect on-premises networks and other VPCs to AWS with Site-to-Site VPN, Direct Connect, VPC peering, Transit Gateway, and VPC endpoints.",16,[24,25,26,27],"Explain how AWS Site-to-Site VPN connects an on-premises network to a VPC","Describe AWS Direct Connect and when its dedicated connection is worth the cost","Compare VPN and Direct Connect across cost, setup, and performance","Describe VPC peering, Transit Gateway, and VPC endpoints for private connectivity",{"locked":10,"reason":3,"meta":29,"item":37},{"title":30,"description":31,"isFree":5,"estimatedMinutes":11,"difficulty":12,"learningObjectives":32},"AWS AI and Machine Learning Services","Learn the AWS AI and ML stack: pre-trained AI services like Rekognition and Comprehend, the Amazon SageMaker platform for building your own models, and generative AI with Amazon Bedrock and Amazon Q.",[33,34,35,36],"Describe the three layers of the AWS AI and ML stack","Match common pre-trained AI services to the problems they solve","Explain what Amazon SageMaker is and how it differs from the pre-trained AI services","Identify Amazon Bedrock and Amazon Q as the AWS generative AI services",{"id":38,"title":30,"body":39,"description":31,"difficulty":12,"estimatedMinutes":11,"extension":327,"infographics":328,"isFree":5,"learningObjectives":340,"meta":341,"navigation":5,"path":342,"quiz":343,"seo":391,"stem":392,"__hash__":393},"courses/courses/aws-certified-cloud-practitioner/en/domains/03-cloud-technology-and-services/07-ai-ml-and-analytics/01-ai-ml-services.md",{"type":40,"value":41,"toc":316},"minimark",[42,47,51,54,58,61,66,73,79,85,88,91,94,238,241,263,267,270,273,276,280,283,289,295,299],[43,44,46],"h2",{"id":45},"why-this-matters","Why this matters",[48,49,50],"p",{},"AWS gives you machine learning at three very different levels of effort. You can call a ready-made AI service and get a result in one API request, you can build and train your own model on the Amazon SageMaker platform, or you can tap into generative AI with foundation models. The CLF-C02 exam does not ask you to train a model or write code. It asks you to recognize the main services and match a described problem to the right one.",[48,52,53],{},"This lesson maps out the AWS AI and ML stack so the names stop blurring together. You will see how the layers differ, what each pre-trained service is for, where SageMaker fits, and which services cover generative AI.",[43,55,57],{"id":56},"the-three-layers-of-the-aws-ai-stack","The three layers of the AWS AI stack",[48,59,60],{},"Think of AWS machine learning as three stacked layers, from least effort to most control.",[62,63],"infographic",{"alt":64,"slug":65},"A diagram of the three layers of the AWS AI and ML stack: pre-trained AI services at the top, the Amazon SageMaker platform in the middle, and generative AI with Amazon Bedrock and Amazon Q at the bottom.","ai-ml-stack-three-layers",[48,67,68,72],{},[69,70,71],"strong",{},"Pre-trained AI services"," sit at the top. AWS already trained these models, so you send data to an API and get an answer back: a label for an image, the sentiment of a sentence, an audio file from text. No ML knowledge is needed. This is where most Cloud Practitioner questions land.",[48,74,75,78],{},[69,76,77],{},"The ML platform",", Amazon SageMaker, sits in the middle. You reach for it when the pre-trained services do not fit and you need to build, train, and deploy your own model on your own data. It still removes a lot of heavy lifting, but you are doing real ML work here.",[48,80,81,84],{},[69,82,83],{},"Generative AI"," is the newest layer. Amazon Bedrock and Amazon Q let you use foundation models to generate text, answer questions, and build assistants, without training a model from scratch.",[48,86,87],{},"A quick rule of thumb: if AWS already solved the problem, use a pre-trained service; if you need a custom model, use SageMaker; if you want generative AI, use Bedrock or Q.",[43,89,71],{"id":90},"pre-trained-ai-services",[48,92,93],{},"These services deliver ready-made intelligence through an API. You do not train anything. The exam expects you to know what each one does and to pick the right one for a scenario.",[95,96,97,113],"table",{},[98,99,100],"thead",{},[101,102,103,107,110],"tr",{},[104,105,106],"th",{},"Service",[104,108,109],{},"What it does",[104,111,112],{},"Typical use",[114,115,116,128,139,150,161,172,183,194,205,216,227],"tbody",{},[101,117,118,122,125],{},[119,120,121],"td",{},"Amazon Rekognition",[119,123,124],{},"Image and video analysis",[119,126,127],{},"Detect objects, people, text, and scenes; facial analysis and comparison; content moderation",[101,129,130,133,136],{},[119,131,132],{},"Amazon Comprehend",[119,134,135],{},"Natural language processing",[119,137,138],{},"Find sentiment, key phrases, entities, and language in text",[101,140,141,144,147],{},[119,142,143],{},"Amazon Textract",[119,145,146],{},"Extract text and data from documents",[119,148,149],{},"Pull text, tables, and form fields from scanned PDFs and images",[101,151,152,155,158],{},[119,153,154],{},"Amazon Translate",[119,156,157],{},"Language translation",[119,159,160],{},"Translate text between languages for websites and apps",[101,162,163,166,169],{},[119,164,165],{},"Amazon Polly",[119,167,168],{},"Text to speech",[119,170,171],{},"Turn written text into lifelike spoken audio",[101,173,174,177,180],{},[119,175,176],{},"Amazon Transcribe",[119,178,179],{},"Speech to text",[119,181,182],{},"Convert spoken audio into written text, such as subtitles or call transcripts",[101,184,185,188,191],{},[119,186,187],{},"Amazon Lex",[119,189,190],{},"Conversational interfaces",[119,192,193],{},"Build text and voice chatbots and IVR systems",[101,195,196,199,202],{},[119,197,198],{},"Amazon Kendra",[119,200,201],{},"Intelligent search",[119,203,204],{},"Search across enterprise documents and get direct answers",[101,206,207,210,213],{},[119,208,209],{},"Amazon Personalize",[119,211,212],{},"Recommendations",[119,214,215],{},"Generate individualized product and content recommendations",[101,217,218,221,224],{},[119,219,220],{},"Amazon Forecast",[119,222,223],{},"Time-series forecasting",[119,225,226],{},"Predict future values like demand or inventory from historical data",[101,228,229,232,235],{},[119,230,231],{},"Amazon Fraud Detector",[119,233,234],{},"Fraud detection",[119,236,237],{},"Spot potentially fraudulent online activity",[48,239,240],{},"A few pairs trip people up, so anchor them now:",[242,243,244,251,257],"ul",{},[245,246,247,250],"li",{},[69,248,249],{},"Polly versus Transcribe."," Polly is text to speech (text goes in, audio comes out). Transcribe is speech to text (audio goes in, text comes out). They are mirror images.",[245,252,253,256],{},[69,254,255],{},"Comprehend versus Translate."," Comprehend analyzes what text means (sentiment, entities, language). Translate converts text from one language to another.",[245,258,259,262],{},[69,260,261],{},"Rekognition versus Textract."," Rekognition analyzes images and video for objects, people, and scenes. Textract pulls structured text and data out of documents like forms and tables.",[43,264,266],{"id":265},"amazon-sagemaker","Amazon SageMaker",[48,268,269],{},"Amazon SageMaker is the platform for building, training, and deploying your own machine learning models. You use it when the pre-trained services do not cover your problem and you need a model trained on your specific data, such as predicting equipment failure from your own sensor readings.",[48,271,272],{},"SageMaker provides the tools for the whole ML workflow in one place: preparing data, picking an algorithm, training the model, tuning it, and deploying it to make predictions. It manages the underlying infrastructure so you focus on the model rather than the servers.",[48,274,275],{},"For the exam, hold one clear line: a pre-trained AI service is a finished product you call through an API, while SageMaker is the workshop where you build a model yourself. If a question describes a team building and training a custom model, the answer is SageMaker.",[43,277,279],{"id":278},"generative-ai-amazon-bedrock-and-amazon-q","Generative AI: Amazon Bedrock and Amazon Q",[48,281,282],{},"Generative AI uses large foundation models to produce new content, such as text, summaries, and answers. AWS offers two services you should recognize.",[48,284,285,288],{},[69,286,287],{},"Amazon Bedrock"," is a fully managed service that makes foundation models from Amazon and leading AI companies available through a single API. You can experiment with these models, customize them with your own data, and build generative AI applications without managing any infrastructure or training a model from scratch. Bedrock is the service for developers building generative AI features into their own apps.",[48,290,291,294],{},[69,292,293],{},"Amazon Q"," is a generative AI assistant. Amazon Q Business answers questions, summarizes information, and completes tasks using your company's internal data, which helps employees work faster. Amazon Q Developer helps with coding, testing, debugging, and working with AWS resources. Think of Q as a ready-to-use assistant, where Bedrock is the building block for your own generative AI apps.",[43,296,298],{"id":297},"exam-tips","Exam tips",[242,300,301,304,307,310,313],{},[245,302,303],{},"The AWS AI stack has three layers: pre-trained AI services (call an API), Amazon SageMaker (build your own model), and generative AI (Amazon Bedrock and Amazon Q).",[245,305,306],{},"Pre-trained AI services need no ML expertise. Match the service to the task: Rekognition (images and video), Comprehend (text meaning), Textract (data from documents), Translate (languages), Polly (text to speech), Transcribe (speech to text), Lex (chatbots), Kendra (search), Personalize (recommendations), Forecast (time-series prediction), Fraud Detector (fraud).",[245,308,309],{},"Remember the mirror pairs: Polly is text to speech, Transcribe is speech to text.",[245,311,312],{},"Amazon SageMaker is the platform to build, train, and deploy custom models. Pick it when a scenario involves building a model on your own data.",[245,314,315],{},"Amazon Bedrock gives access to foundation models through one API for building generative AI apps. Amazon Q is a ready-to-use generative AI assistant.",{"title":317,"searchDepth":318,"depth":318,"links":319},"",3,[320,322,323,324,325,326],{"id":45,"depth":321,"text":46},2,{"id":56,"depth":321,"text":57},{"id":90,"depth":321,"text":71},{"id":265,"depth":321,"text":266},{"id":278,"depth":321,"text":279},{"id":297,"depth":321,"text":298},"md",[329],{"slug":65,"concept":330,"style":331,"aspectRatio":332,"labels":333},"A vertical stack of the three AWS machine learning layers with an effort-versus-control axis running down its side. Pre-trained AI services sit at the top, called through an API with no ML work; Amazon SageMaker sits in the middle for building and training your own models on your own data; the generative AI layer sits at the base with Amazon Bedrock as the building block for your own apps and Amazon Q as the ready-to-use assistant, both using foundation models with no training from scratch. The orange accent highlights the top pre-trained layer, where most Cloud Practitioner questions land. Takeaway: less effort as you move up the stack, more control as you move down.","diagram","16:9",[334,335,336,337,338,339],"Pre-trained services: call an API, no ML","SageMaker: build and train your own model","Generative AI: foundation models, no training","Bedrock: building block for your own apps","Amazon Q: ready-to-use assistant","Less effort above, more control below",[33,34,35,36],{},"/courses/aws-certified-cloud-practitioner/en/domains/03-cloud-technology-and-services/07-ai-ml-and-analytics/01-ai-ml-services",{"passingScore":344,"questions":345},70,[346,355,359,365,369,373,381,387],{"question":347,"type":348,"options":349,"correctAnswer":350,"explanation":354},"What is the main difference between the AWS pre-trained AI services (like Amazon Rekognition) and Amazon SageMaker?","single",[350,351,352,353],"Pre-trained AI services work through an API with no model building, while SageMaker is a platform to build, train, and deploy your own models","Pre-trained AI services cost more because they include support, while SageMaker is always free","Pre-trained AI services run only on-premises, while SageMaker runs only in the cloud","There is no difference; they are different names for the same product","The pre-trained AI services give you ready-made intelligence through a simple API call, with no ML expertise required. Amazon SageMaker sits one layer down: it is the platform you use when you need to build, train, and deploy a custom model on your own data.",{"question":356,"type":348,"options":357,"correctAnswer":176,"explanation":358},"A media company wants to add subtitles to its videos by automatically converting the spoken audio into text. Which AWS service fits best?",[165,176,154,132],"Amazon Transcribe is the speech-to-text service, so it turns spoken audio into written text for subtitles. Polly does the reverse (text to speech), Translate converts between languages, and Comprehend analyzes the meaning of text.",{"question":360,"type":361,"options":362,"correctAnswers":363,"explanation":364},"Which of the following are pre-trained AI services that you can use through an API without any ML expertise? (Select all that apply.)","multiple",[121,132,187,266],[121,132,187],"Rekognition, Comprehend, and Lex are pre-trained AI services that work through an API with no model building required. SageMaker is the build-your-own-model platform, which is a different layer of the stack.",{"question":366,"type":348,"options":367,"correctAnswer":165,"explanation":368},"Which AWS service turns written text into lifelike speech?",[176,143,165,198],"Amazon Polly is the text-to-speech service that produces lifelike spoken audio from text. Transcribe goes the other way (speech to text), Textract extracts text from scanned documents, and Kendra is an intelligent search service.",{"question":370,"type":348,"options":371,"correctAnswer":187,"explanation":372},"A developer wants to build a text and voice chatbot to handle customer questions. Which service is purpose-built for this?",[187,121,220,154],"Amazon Lex builds conversational interfaces (chatbots) using the same speech recognition and natural language understanding that powers Alexa. Rekognition handles images and video, Forecast predicts time-series data, and Translate converts languages.",{"question":374,"type":348,"options":375,"correctAnswer":377,"explanation":380},"What does Amazon Bedrock provide?",[376,377,378,379],"A 1/18th scale race car for learning reinforcement learning","Access to foundation models from Amazon and other AI companies through a single API, for building generative AI applications","A service that extracts text from scanned PDFs and forms","A managed data warehouse for business intelligence","Amazon Bedrock is the managed generative AI service that makes foundation models available through one API, so you can build generative AI apps without managing the underlying infrastructure. The race car describes AWS DeepRacer, and the other options describe Textract and Amazon Redshift.",{"question":382,"type":348,"options":383,"correctAnswer":384,"explanation":386},"Amazon Rekognition is the right service to identify objects, people, and text in images and video.",[384,385],"True","False","Amazon Rekognition adds image and video analysis to your applications: it can identify objects, people, text, and scenes, and run facial analysis and comparison. It requires no ML expertise to use.",{"question":388,"type":348,"options":389,"correctAnswer":209,"explanation":390},"A retailer wants to show each shopper individualized product recommendations based on their browsing and purchase history. Which AWS service is built for this?",[198,132,209,165],"Amazon Personalize creates individualized recommendations using the same technology behind recommendations on Amazon.com. Kendra is enterprise search, Comprehend analyzes text, and Polly produces speech.",{"title":30,"description":31},"courses/aws-certified-cloud-practitioner/en/domains/03-cloud-technology-and-services/07-ai-ml-and-analytics/01-ai-ml-services","Cpn7E4eIOrI6HDOOm8qNpf4lz5lhk3N3J650vkLH3Lc"]