[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"lesson-aws-certified-ai-practitioner-genai-fundamentals-en":3,"cheat-sheet---en":3,"topic-info----en":3,"prev-aws-certified-ai-practitioner-genai-fundamentals-en":3,"next-aws-certified-ai-practitioner-genai-fundamentals-en":3,"domain-info-aws-certified-ai-practitioner-genai-fundamentals-en":4},null,{"meta":5,"body":8},{"title":6,"description":7},"Fundamentals of Generative AI","Foundation models, tokens, embeddings, and agentic AI: how generative AI works, what it costs, and which AWS services bring it to production.",{"type":9,"value":10,"toc":58},"minimark",[11,15,18,23,48,52,55],[12,13,14],"p",{},"Most people meet generative AI through a chat window and never see what is happening behind it. That gap becomes a problem the moment you have to pick a model, estimate a monthly bill, or explain to a stakeholder why the same question sometimes gets a different answer. This domain closes the gap.",[12,16,17],{},"It starts with the mechanics: what a foundation model is, how text becomes tokens and embeddings, and how diffusion and multimodal models differ from language models. From there it moves to what generative AI is genuinely good at, where it fails, and how a team decides whether the value is worth the spend. The domain ends on AWS ground, with the services that turn a model into a running product.",[19,20,22],"h2",{"id":21},"what-this-domain-covers","What This Domain Covers",[24,25,26,30,33,36,39,42,45],"ul",{},[27,28,29],"li",{},"foundation models and large language models, and how they differ from traditional ML models",[27,31,32],{},"tokens, chunking, and embeddings, plus the token-based pricing that follows from them",[27,34,35],{},"multimodal and diffusion models, and the generative AI use cases each one serves",[27,37,38],{},"the foundation model lifecycle, context engineering, and agentic AI with MCP",[27,40,41],{},"what generative AI does well, its limitations and risks, and how to select a model",[27,43,44],{},"measuring business value against real cost tradeoffs",[27,46,47],{},"the AWS stack: Amazon Bedrock, SageMaker AI and JumpStart, Amazon Q, Kiro, and the agentic tooling around them",[19,49,51],{"id":50},"why-it-matters","Why It Matters",[12,53,54],{},"This is the second-largest domain by exam weight, and it feeds directly into the largest one. Every question about RAG, prompt engineering, or fine-tuning assumes you already know what a token is, why context has a limit, and what Bedrock actually provides.",[12,56,57],{},"The practical payoff is judgment. Once you can trace a request from prompt to tokens to inference to invoice, you can answer the questions that matter on the job: is this task a good fit for a foundation model, which model should we start with, and what will it cost us at scale.",{"title":59,"searchDepth":60,"depth":60,"links":61},"",3,[62,64],{"id":21,"depth":63,"text":22},2,{"id":50,"depth":63,"text":51}]