▣ Chapter 1: Basics and Trends of Generative AI 1.1 What is 'Generative AI'? __1.1.1 The Position of Artificial Intelligence (AI) and Generative AI __1.1.2 What is a 'Model' in Generative AI __1.1.3 Basic Knowledge about 'Models' 1.2 Famous Generative AI Products __1.2.1 ChatGPT __1.2.2 Stable Diffusion __1.2.3 GitHub Copilot 1.3 Providing APIs for Generative AI and Deploying them in the Cloud __1.3.1 APIs for Generative AI Models __1.3.2 APIs for Generative AI Models Provided in the Cloud ▣ Chapter 2: Introduction to Amazon Bedrock 2.1 What is Amazon Bedrock? __2.1.1 Advantages of Bedrock __2.1.2 Supported AWS Regions __2.1.3 Bedrock Model Usage Fees 2.2 Why Choose AWS's Bedrock? __2.2.1 (1) Most of AWS's strengths can be utilized __2.2.2 (2) Wide availability of the latest models provided by various companies __2.2.3 (3) High convenience of application development __2.2.4 (4) Enterprise-level security and governance provided 2.3 Generative AI models available in Bedrock __2.3.1 Types of models __2.3.2 Bedrock's recommended models 2.4 Anthropic's generative AI model __2.4.1 Characteristics of Anthropic's model __2.4.2 Claude 3 series 2.5 Cohere's generative AI model __2.5.1 Cohere's model characteristics __2.5.2 Command R series __2.5.3 Embed English / Multilingual 2.6 Stability AI's generative AI model __2.6.1 Stability AI's model characteristics __2.6.2 Stable Diffusion 3.5 Large 2.7 Amazon's Generative AI Models __2.7.1 Characteristics of Amazon Models __2.7.2 Amazon Nova Recognition Model __2.7.3 Amazon Nova Creative Content Generation Model 2.8 Meta's Generative AI Model __2.8.1 Characteristics of Meta Model __2.8.2 Llama 3.3 2.9 Mistral AI's Generative AI Model __2.9.1 Characteristics of Mistral AI's Model __2.9.2 Mistral Large2 / Small 2.10 AI21 Labs' Generative AI Model __2.10.1 Characteristics of AI21 Labs' Model __2.10.2 Jamba 1.5 Large 2.11 [Hands-on] Using Bedrock in Practice __2.11.1 How to Create in a GUI Environment through Playground __2.11.2 How to Send Each Model API Request Using the AWS SDK ▣ Chapter 03: How to Develop Generative AI Applications 3.1 What is a Prompt __3.1.1 How to Write a Prompt __3.1.2 Types of prompts 3.2 What is a token? __3.2.1 Splitting a string into tokens __3.2.2 How to count the number of tokens 3.3 What is prompt engineering? __3.3.1 Guidelines for prompt engineering __3.3.2 Activating a model __3.3.3 Setting a clear task __3.3.4 Providing documentation __3.3.5 Setting specific instructions __3.3.6 Providing examples __3.3.7 Inducing step-by-step thinking __3.3.8 Other prompt engineering techniques 3.4 Major frameworks used to develop generative AI apps __3.4.1 Utilizing generative AI frameworks __3.4.2 LangChain __3.4.3 Streamlit 3.5 Developing generative AI apps using LangChain and Streamlit __3.5.1 Preparing the development environment __3.5.2 [Step 1] Implementing LangChain __3.5.3 [Step 2] Outputting a stream __3.5.4 [Step 3] Integrating Streamlit __3.5.5 [Step 4] Implementing continuous chat conversations __3.5.6 [Step 5] Maintaining chat history 3.6 Developing a generative AI app that runs on AWS Lambda __3.6.1 Generative AI app using AWS Lambda __3.6.2 Use cases __3.6.3 Configuring the development environment __3.6.4 Implementation details __3.6.5 Creating a Lambda layer __3.6.6 Creating a Lambda function 3.7 Other frameworks used to develop generative AI apps __3.7.1 LlamaIndex __3.7.2 Gradio __3.7.3 Chainlit __3.7.4 Dify __3.7.5 LiteLLM ▣ Chapter 04: Let's create an in-house document search RAG application 4.1 What is RAG? __4.1.1 Characteristics and use cases of RAG __4.1.2 'Embedding' that enables semantic search __4.1.3 Implementation example of RAG architecture 4.2 [Hands-on] Let's implement RAG with knowledge base __4.2.1 Structure of knowledge base __4.2.2 Overview of RAG application development using knowledge base __4.2.3 Creating an S3 bucket __4.2.4 Creating a knowledge base __4.2.5 Activating a model __4.2.6 Checking the knowledge base's independent operation __4.2.7 Implementing a front-end __4.2.8 Running a RAG application __4.2.9 How to delete unnecessary resources __4.2.10 Generative AI model supporting the knowledge base __4.2.11 Setting up queries for the knowledge base __4.2.12 Usage fees for the knowledge base 4.3 Introduction to search target services for RAG __4.3.1 List of services introduced in this section __4.3.2 Amazon OpenSearch Service (Vector DB/AWS Service) __4.3.3 Amazon OpenSearch Serverless (Vector DB/AWS Service) __4.3.4 Amazon Aurora & Amazon RDS (Vector DB/AWS Service) __4.3.5 Amazon DocumentDB (Vector DB/AWS Service) __4.3.6 Amazon MemoryDB for Redis (Vector DB/AWS Service) __4.3.7 Pinecone (Vector DB/AWS Marketplace Product) __4.3.8 Redis Enterprise Cloud (Vector DB/AWS Marketplace Product) __4.3.9 MongoDB Atlas (Vector DB/AWS Marketplace Product) __4.3.10 Amazon Kendra (Other/AWS Service) __4.3.11 Amazon DynamoDB (Other/AWS Service) __4.3.12 Amazon S3 (Other/AWS Service) 4.4 Recommended RAG Architecture Example __4.4.1 Try it out & Low-cost operation __4.4.2 Answer quality 4.4.3 Emphasis on connectivity with data sources 4.5 How to improve the response quality of RAG __4.5.1 Adjusting chunk size __4.5.2 Adding metadata __4.5.3 Rerank __4.5.4 RAG fusion __4.5.5 Rewrite-Retrieve-Read __4.5.6 HyDE（Hypothetical Document Embeddings） __4.5.7 Other new methods 4.6 Evaluation tools for RAG applications __4.6.1 Ragas __4.6.2 LangSmith __4.6.3 Langfuse ▣ Chapter 05: Creating a convenient autonomous AI agent 5.1 What is an AI agent __5.1.1 AI agent using tools __5.1.2 What is 'ReAct', an advanced AI agent implementation method? __5.1.3 Open source AI agent __5.1.4 Use cases of AI agents 5.2 [Hands-on] Implementing AI agents in LangChain __5.2.1 Preparation __5.2.2 Hands-on ① AI agents using tools __5.2.3 Hands-on ② ReAct agent 5.3 What is Agents for Amazon Bedrock? __5.3.1 Overview of Agents for Amazon Bedrock __5.3.2 Structure of Agents __5.3.3 Details of Agents __5.3.4 Supported models and regions __5.3.5 Agents usage fees 5.4 [Hands-on] Let's create an AI agent with Agents __5.4.1 Overview of the AI ​​agent developed in this chapter __5.4.2 Activating the model __5.4.3 Preparing Pinecone __5.4.4 Creating an S3 bucket __5.4.5 Creating a knowledge base __5.4.6 Creating a Lambda layer __5.4.7 Create Agents __5.4.8 Add a Workgroup __5.4.9 Configure a Lambda Function __5.4.10 Add a Knowledge Base __5.4.11 Create an Alias ​​__5.4.12 Verify the Action __5.4.13 Show Traces __5.4.14 Change the Orchestration Strategy ▣ Chapter 6: Using Bedrock Features 6.1 Customizing Models __6.1.1 What is a Custom Model __6.1.2 Fine-tuning __6.1.3 Continuous Pre-training __6.1.4 Importing a Custom Model 6.2 Safeguards __6.2.1 What is a Safeguard __6.2.2 Watermark Detection __6.2.3 Guardrails 6.3 Evaluation and Deployment __6.3.1 Model Evaluation __6.3.2 Provisioned Throughput 6.4 Other Bedrock Features __6.4.1 Batch Inference __6.4.2 Amazon Bedrock in SageMaker Unified Studio ▣ Chapter 7: Integration of Bedrock with various AWS services 7.1 Integration with Amazon CloudWatch __7.1.1 CloudWatch overview __7.1.2 CloudWatch Metrics __7.1.3 CloudWatch Logs 7.2 Integration with AWS CloudTrail __7.2.1 CloudTrail overview __7.2.2 Management events and data events 7.3 Integration with AWS PrivateLink __7.3.1 PrivateLink overview __7.3.2 Network design for generative AI apps 7.4 Integration with AWS CloudFormation __7.4.1 CloudFormation overview 7.5 Integration with other AWS services __7.5.1 Amazon Aurora __7.5.2 Amazon CodeCatalyst __7.5.3 Amazon Lex __7.5.4 Amazon Transcribe __7.5.5 Amazon Connect ▣ Chapter 8: Let's develop a generative AI app with low code 8.1 AWS Step Functions and prompts Chaining __8.1.1 What is Step Functions __8.1.2 Types of integration __8.1.3 What is prompt chaining __8.1.4 How to use Workflow Studio __8.1.5 Reference values ​​and built-in functions using JSONata syntax 8.2 [Hands-on] Developing a generative AI app using Bedrock and Step Functions __8.2.1 Preparing the development environment __8.2.2 Hands-on overview __8.2.3 【Step 1】 Importing posts about Bedrock __8.2.4 【Step 2】 Summarizing the main points of each imported post __8.2.5 Running the created task test __8.2.6 【Step 3】 Writing a self-introduction and catchphrase __8.2.7 【Step 4】 Converting the created content to Markdown format __8.2.8 [Step 5] Creating a thumbnail image __8.2.9 Running the completed state machine __8.2.10 Creating a confirmation screen ▣ Chapter 9: Introduction to Generative AI-related services other than Bedrock 9.1 AWS's Generative AI Stack __9.1.1 Types of AWS Generative AI Stack __9.1.2 Amazon Q 9.2 If you want to use generative AI as an application __9.2.1 PartyRock __9.2.2 AWS HealthScribe 9.3 If you need training and inference infrastructure for generative AI models __9.3.1 Amazon SageMaker __9.3.2 Amazon SageMaker JumpStart __9.3.3 Amazon SageMaker Canvas __9.3.4 AWS's own designed chips 9.4 [Hands-on] Developing an Amazon Q Business Application __9.4.1 Preparing documents to use in RAG __9.4.2 Creating an AWS IAM Identity Center __9.4.3 Creating a user to log in to the Amazon Q Business application __9.4.4 Creating an Amazon Q Business application __9.4.5 Verifying the operation of the Amazon Q Business application __9.4.6 Setting Amazon Q Business application options __9.4.7 Hands-on environment deletion ▣ Chapter 10: Bedrock use cases 10.1 Miridi case __10.1.1 Introduction to Miridi __10.1.2 AI presentation creation function __10.1.3 Architecture of Miridi Canvas __10.1.4 Service performance and future plans 10.2 Today's House case __10.2.1 Introduction to Today's House __10.2.2 Introduction to the Oh Butler project __10.2.3 Architecture of Oh Butler __10.2.4 Service performance and future plans 10.3 AgentSoft (Happy Campus) case __10.3.1 Introduction to AgentSoft __10.3.2 Introduction to EasyAI service __10.3.3 Introduction to EasyAI functions __10.3.4 EasyAI architecture __10.3.5 Service performance and future plans ▣ Chapter 11: Catch up on the latest information 11.1 AWS official materials __11.1.1 Official documentation __11.1.2 AWS What's New __11.1.3 AWS Blog __11.1.4 GitHub Public Data __11.1.5 AWS Innovate 11.2 Technical Community and Information Collection Platform __11.2.1 AWSKRUG (AWS Korea User Group) __11.2.2 Qiita (Japanese Developer Site) __11.2.3 X (formerly Twitter) __11.2.4 Discord __11.2.5 LinkedIn ▣ Appendix 1: AWS Account Creation Procedure Creating an AWS Account Setting Up MFA (Multi-Factor Authentication) ▣ Appendix 2: IAM User Creation Procedure Creating a New IAM User Creating MFA (Multi-Factor Authentication) ▣ Appendix 3: Hands-on Environment Configuration