[Generative Agents Co., Ltd.] Practical guide for AI agent development “Introduction to RAG/AI agent [practice] using LangChain and LangGraph” will be released on November 9th #AI agent practice book
Generative Agents Co., Ltd. Press release: November 13, 2024 Practical guide for AI agent development “Introduction to RAG/AI agent [Practical] using LangChain and LangGraph” will be released on November 9th #AI agent practice book From search extension generation (RAG) to design patterns in AI agent development, a manual covering everything from basic knowledge to practical hands-on is now available! This book will help you understand the trends in LLM app development.
https://prcdn.freetls.fastly.net/release_image/147692/4/147692-4-8fceb150d00c67fa06ca87839f164b48-3900×2410.png
978-4-297-14530-9 https://gihyo.jp/book/2024/978-4-297-14530-9 In addition, we will also begin offering LLM application developer training programs for companies that utilize the knowledge contained in this book, as well as consulting services. ▼Display status at bookstores Co-author Shingo Yoshida visited bookstores and placed POP cards handwritten by the author. Maruzen Marunouchi Main Store Image
URL: https://prcdn.freetls.fastly.net/release_image/147692/4/147692-4-924aff8720fa2b9ea8e2767c87a55586-1536×2048.jpg Maruzen Marunouchi main store 3rd floor computer corner
https://prcdn.freetls.fastly.net/release_image/147692/4/147692-4-c4e9cb391e90f8595d6affe40ebcf2ef-1536×2048.jpg Maruzen Marunouchi main store 3rd floor computer corner Junkudo Bookstore Ikebukuro Main Store
https://prcdn.freetls.fastly.net/release_image/147692/4/147692-4-b2c8f301f24d7a9bceb92b51c79331dc-1536×2048.jpg Junkudo Bookstore Ikebukuro Main Store 6th floor computer corner Image
URL: https://prcdn.freetls.fastly.net/release_image/147692/4/147692-4-0bef4e6255fe7a223ced103afc52d1ba-1536×2048.jpg Junkudo Bookstore Ikebukuro Main Store 6th floor computer corner Kinokuniya Bookstore Shinjuku Main Store
https://prcdn.freetls.fastly.net/release_image/147692/4/147692-4-b6ad11ec303e92ba1172805c17c145e1-1536×2048.jpg Kinokuniya Bookstore Shinjuku Main Store 5th floor computer book Image
URL: https://prcdn.freetls.fastly.net/release_image/147692/4/147692-4-56696bae9f877009ecefcaee5440be20-1536×2048.jpg Kinokuniya Bookstore Shinjuku Main Store 5th floor computer book ▼Overview and main features of this book This book is a practical guide to developing applications using LLM (Large-Scale Language Models), systematically covering the following three areas: 1. Basics and practice of LLM utilization • Explanation from the basics to details of using OpenAI’s chat API • Prompt engineering techniques necessary for LLM application development • Basics of application development using LangChain 2. Implementation and evaluation of RAG (Search Extension Generation) • Devising search queries, utilizing rerank models and hybrid searches • How to implement Advanced RAG • Evaluation method for RAG applications using LangSmith 3. Fundamentals and practice of AI agent development • Origin and evolution of AI agents using LLM • Basics and practice of AI agent development using LangGraph • Explanation of 18 types of agent design patterns • Implementation explanation of agent design pattern using
LangChain/LangGraph Since the advent of ChatGPT, the trend of incorporating generative AI such as LLM into all applications has accelerated, but the basic information source is English, and systematic information provision in Japanese has not progressed yet. That is the reality. In particular, most of the information on the libraries LangChain/LangGraph, which are used as the de facto standard for LLM application development, is provided in English. In response to this situation, the three founding members of our company, who are official LangChain experts, wrote a book called “Introduction to RAG/AI Agent [Practical] Using LangChain and LangGraph.” As you can see from the table of contents, it comprehensively summarizes everything from how to use LangChain and LangGraph to the development method of RAG (Retrieval Augmentation Generation), which is a standard pattern of LLM applications, and the development method of AI agent. Also, even if you are a beginner, if you read carefully from the beginning, you will be able to step up from the basics to AI agent development. This book is a must-have for those who think, “AI agents are amazing! But I don’t know anything about AI, so I don’t know if it’s something I can do.” August 8, 2024 Generative Agents Co., Ltd. is the first Japanese company to be certified as an official LangChain expert ▼Table of contents Chapter 1 Fundamentals of LLM Application Development • 1.1 Generative AI has begun to be used • 1.2 Copilot vs AI agent • 1.3 Everything becomes an AI agent • 1.4 AI agent knowledge map • 1.5 Summary Chapter 2 Basics of OpenAI chat API • 2.1 OpenAI chat model • 2.2 Basics of OpenAI chat API • 2.3 “Tokens” that affect input/output length limits and fees • 2.4 Preparing the environment to try Chat Completions API • 2.5 Hands-on with Chat Completions API • 2.6 Function calling • 2.7 Summary Chapter 3 Prompt Engineering • 3.1 Necessity of prompt engineering • 3.2 What is prompt engineering? • 3.3 Basics of prompt components • 3.4 Standard method of prompt engineering • 3.5 Summary Chapter 4 Basics of LangChain • 4.1 Overview of LangChain • 4.2 LLM/Chat model • 4.3 Prompt template • 4.4 Output parser • 4.5 Overview of Chain―LangChain Expression Language (LCEL) • 4.6 LangChain RAG components • 4.7 Summary Chapter 5 A thorough explanation of LangChain Expression Language (LCEL) • 5.1 Runnable and RunnableSequence – the most basic components of LCEL • 5.2
RunnableLambda – Make any function Runnable • 5.3 RunnableParallel – Connect multiple Runnables in parallel • 5.4 RunnablePassthrough – Output the input as is • 5.5 Summary Chapter 6 Advanced RAG • 6.1 Overview of Advanced RAG • 6.2 Hands-on preparation • 6.3 Search query innovation • 6.4 Post-search tips • 6.5 Ideas for using multiple Retrievers • 6.6 Summary Chapter 7. Evaluating RAG applications using LangSmith • 7.1 Overview of evaluation addressed in Chapter 7 • 7.2 Overview of LangSmith • 7.3 Configuration example for offline evaluation using LangSmith and Ragas • 7.4 Generating synthetic test data with Ragas • 7.5 Implementing offline evaluation using LangSmith and Ragas • 7.6 Gathering feedback using LangSmith • 7.7 Automatic processing for leveraging feedback • 7.8 Summary Chapter 8 What is AI agent • 8.1 Expectations for utilizing LLM for AI agents • 8.2 Origin of AI agent and evolution of AI agent using LLM • 8.3 Generic LLM agent framework • 8.4 Multi-agent approach • 8.5 For the safe spread of AI agents • 8.6 Summary Chapter 9 Introduction to practical AI agent creation with LangGraph • 9.1 Overview of LangGraph • 9.2 Main components of LangGraph • 9.3 Hands-on: Q&A application • 9.4 Checkpointing: State Persistence and Resumption • 9.5 Summary Chapter 10 Development of requirements definition document generation AI agent • 10.1 Overview of requirements definition document generation AI agent • 10.2 Environment settings • 10.3 Defining data structures • 10.4 Implementation of main components • 10.5 Workflow construction • 10.6 Run the agent and check the results • 10.7 Entire source code • 10.8 Summary Chapter 11 Agent Design Pattern • 11.1 Overview of agent design patterns • 11.2 18 Agent Design Patterns • Overall diagram of agent design pattern • Passive Goal Creator • Proactive Goal Creator • Prompt/Response Optimizer • Retrieval-Augmented Generation (RAG) • Single-Path Plan Generator • Multi-Path Plan Generator •
Self-Reflection • Cross-Reflection • Human-Reflection • One-Shot Model Querying • Incremental Model Querying • Voting-Based Cooperation • Role-Based Cooperation • Debate-Based Cooperation • Multimodal Guardrails • Tool/Agent Registry • Agent Adapter • Agent Evaluator • 11.3 Summary Chapter 12 Agent design patterns implemented with LangChain/LangGraph • 12.1 Agent design patterns covered in this chapter • 12.2 Environment settings • 12.3 Passive Goal Creator • 12.4 Prompt/Response Optimizer • 12.5 Single-Path Plan Generator • 12.6 Multi-Path Plan Generator • 12.7 Self-Reflection • 12.8
Cross-Reflection • 12.9 Role-Based Cooperation • 12.10 Summary ▼Author information It is co-authored by three people who are also certified as LangChain ambassadors. https://www.langchain.com/community Kimihiro Nishimi Generative Agents Co., Ltd. Representative Director and CEO While developing applications using large-scale language models and providing advisory services, with a focus on the use of ChatGPT, he met Yoshida and Oshima and co-founded Generative Agents, Inc. By introducing AI agents into management, we are working hard to increase productivity in all industries and business types. “An AI agent has done that job. — The autonomous AI revolution that will come after ChatGPT” (Gijutsu Hyoronsha), and Software Design, “Practical LLM Application Development” (Gijutsu Hyoronsha), serialized. Shingo Yoshida Generative Agents Co., Ltd. Director and COO / Section Nine Co., Ltd. Representative Director and CEO Promoting the spread of serverless in Japan as an AWS Serverless Hero. Co-authored
“Introduction to building a chat system using ChatGPT/LangChain [Practical]” (Gijutsu Hyoronsha), Co-authored “Introduction to ChatGPT/LLM system construction using Azure OpenAI Service” (Gijutsu Hyoronsha), “Serverless architecture using AWS” (Shoeisha) Supervised, translated and supervised “Serverless Single Page Applications” (O’Reilly), co-authored “AWS Expert Training Book” (Gijutsu
Hyoronsha). Sponsored by ChatGPT Community (JP) Yuki Oshima Generative Agents Co., Ltd. Director CTO Developed applications and AI agents incorporating large-scale language models. Personally, I hold study sessions for engineers and create teaching materials. Online courses Udemy has many best-selling courses. Operates study group community StudyCo. Co-author of “Introduction to Chat System Construction [Practical] Using ChatGPT/LangChain” (Gijutsu Hyoronsha). ▼Scheduled publication commemorative event November 14th (Thursday) Sponsored by StudyCo “How to walk through the book “Introduction to RAG/AI agent [Practical] using LangChain and LangGraph” Event date and time: 2024/11/14 (Thursday) 19:30 – 20:30 Event format: Online only (archive will be distributed at a later date) Participation fee: Free https://studyco.connpass.com/event/335350/ Wednesday, December 11th sponsored by WAKE Career “Must-see for female engineers! Introduction to RAG/AI agent [practical] using LangChain and LangGraph” Event date and time: 2024/12/11 (Wed) 19:00 – 20:30 Event format: Online only (archive will be distributed only to WAKE Career registrants at a later date) Participation fee: Free
https://wake-career.connpass.com/event/336681/ ▼In conjunction with the publication of the book, we will also start offering LLM application developer training courses for companies as well as consulting services! Along with the publication of the book, our company plans to start offering the “LLM Application Developer Training Course” (hereinafter referred to as the “Course”) for companies. In this course, we will prepare a “1-day basic course” and a “3-day practical course” for each level as a training program that allows you to efficiently learn everything from the basics of LangChain to practical application development. • In the 1-day basic course, you will learn everything from the basics of prompt
engineering to the implementation of basic chatbots and RAG (search augmentation generation) applications, aiming to reach a level where you can participate in the development of LLM applications as a member. • In the 3-day practical course, you will learn the basics of LLMOps for advanced RAG methods, evaluation, and operation, aiming to reach a level where you can lead the development of LLM applications as a leader. In addition, you will learn about the basics of AI agent development and the design patterns for developing AI agents. As an example, the curriculum structure of a one-day course is as follows. [Curriculum composition] 1. Fundamentals of LLM Application
Development (2 hours) 2. Introduction to LangChain (2 hours) 3. RAG app development using LangChain (2 hours) 4. Q&A (30 minutes) We also have customized plans to suit your needs, so please feel free to contact us. We also provide consulting services to promote LLM application development. Please consider this as well. Inquiries regarding courses ▼Company information Image
URL: https://prcdn.freetls.fastly.net/release_image/147692/4/147692-4-46427c1044168303bef437f82f11eeec-1950×574.png Generative Agents Co., Ltd. Generative Agents Co., Ltd. is a company that works on the development and promotion of the use of AI agents using the latest AI technology, with the theme of “creating a world that collaborates with AI agents.” Main business: Provision of generative AI application development support, consulting, education and training services centered on AI agent technology Date of establishment: March 14, 2024 Head office location: Minato-ku, Tokyo Representative Director: Kimihiro Nishimi Website:
https://www.generative-agents.co.jp/