First Yard | Getting to Know LLM

Chapter 1 What can I do with an LLM?
_01-1 The era of generative AI started with ChatGPT
__Large-Scale Language Model, What is LLM?
__Types of LLM
Types of generative AI services using __LLM
_01-2 Why should I study LLM?
__Why LLM Programming Experience is Needed
__Which language model should I choose?
__6 Technologies to Overcome the Limitations of LLM

Chapter 2 Setting Up the Environment and Getting Started with the GPT API
_02-1 Setting up the Python programming environment
__[Do it! Practice] Installing Python
__[Do it! Practice] Installing Visual Studio Code
__[Do it! Practice] Creating a Virtual Environment
_02-2 Getting Started with the GPT API
__What is API?
__OpenAI's API
__[Do it! Practice] Obtaining an OpenAI API Key
__[Do it! Practice] Ask and Get Answers with OpenAI's API Key
__[Do it! Practice] Managing API Keys

Chapter 3: Building a Chatbot with OpenAI's API
_03-1 Learn about Prompt Engineering
__[Do it! Practice] Assigning Roles to GPT
__[Do it! Practice] Applying One-Shot and Few-Shot Prompting
_03-2 Multi-turn conversation with GPT
__[Do it! Practice] Creating a Chatbot with Multi-Turn Conversations
_03-3 Completing a Chatbot with Streamlet
__[Do it! Practice] Creating a Chatbot UI with Streamlet

Second Yard | Task Automation Using OpenAI's GPT API

Chapter 4: AI Researchers Summarizing Documents and Papers
_04-1 Preprocessing PDF documents
__[Do it! Practice] Converting PDF Files to Text Files
__[Do it! Practice] Preprocessing PDF Files
_04-2 Complete the AI ​​Researcher's Summary of the Paper
__[Do it! Practice] Creating a Text Summary Prompt
__[Do it! Practice] Summarizing and printing PDF content

Chapter 5: AI Secretary Organizing Meeting Minutes
_05-1 Converting voice to text
__[Do it! Practice] Using the Whisper API
_05-2 Converting speech to text locally
__Hugging Face
__[Do it! Practice] Downloading the Whisper Model and Using It Locally
_05-3 Distinguishing between sentences and speakers
__[Do it! Practice] Distinguishing speakers by time zone using the speaker separation model
__[Do it! Practice] Analyzing Sentences and Matching Speakers with Pandas
_05-4 Completing the AI ​​Secretary Organizing Meeting Minutes
__[Do it! Practice] Summarizing the entire meeting
__[Do it! Practice] Proofreading Transcripts with GPT

Chapter 6: AI Image Analyst Using GPT-4o
_06-1 Asking GPT Vision to describe an image
__[Do it! Practice] Requesting explanations using images from the internet
__[Do it! Practice] Requesting a description of an image I have
__[Do it! Practice] Understanding the Limitations of GPT Vision
_06-2 Create a quiz using images
__[Do it! Practice] Creating a Problem-Generating Function
__[Do it! Practice] Creating Problems in English
__[Do it! Practice] Creating English Listening Assessment Questions with TTS

Chapter 7: AI Investor Provides Latest Stock Information
_07-1 Basics of Function Calling
__GPT, what time is it now?
__What is function calling?
__[Do it! Practice] Applying Function Calling
__What time is it in New York now?
__[Do it! Practice] Telling the time in each city
__[Do it! Practice] Answering the times in multiple cities at once
__[Do it! Practice] Using Function Calling in Streamlet
_07-2 Talking about GPT and US stocks
__[Do it! Practice] Using yfinance
__[Do it! Practice] Creating yfinance-related functions for use in GPT
__[Do it! Practice] Refactoring Code
__[Do it! Practice] Get the latest stock price and recommendation information for a stock
_07-3 Outputting the stream 180
__[Do it! Practice] Streaming output in the terminal window
__[Do it! Practice] Streaming Output in Streamlet
__[Do it! Practice] Using Function Calling in Stream Mode

Third Yard | Agent Development Using Langchain

Chapter 8: Creating an Agent with Langchain
_08-1 Creating a Chatbot with Langchain
__What is a lang chain?
__[Do it! Practice] Comparing Langchain and OpenAI's GPT API
__[Do it! Practice] Multi-turn conversation with Langchain
_08-2 Creating a chain with LCEL
__[Do it! Practice] Output Parser and Chain
__[Do it! Practice] Using Prompt Templates
_08-3 Creating an Agent with Langchain Tools
__[Do it! Practice] Connecting Functions to a Lang Chain with the @tool Decorator
__[Do it! Practice] Using Pidantik
_08-4 Outputting in Stream Mode
__[Do it! Practice] Outputting a Stream When Using a Tool
_08-5 Implementing in Streamlet
__[Do it! Practice] Building a Multi-Turn Chatbot Based on Langchain Memory
__[Do it! Practice] Creating a Multiturn Without Langchain Memory
__[Do it! Practice] Adding Tools and Streaming Output

Chapter 9: Creating a Document-Based Chatbot with RAG
_09-1 What is RAG?
__How Language Models and RAG Work
__Difference between the basic language model's response and RAG's
__Chunking: Splitting large documents into smaller chunks
__Embedding: Converting Text to Vectors
__Vector DB and Retriever
__Query Extension
_09-2 Implementing a Chatbot Based on RAG
__[Do it! Practice] Converting PDF files to text and splitting them into chunks
__[Do it! Practice] Using the OpenAI Embedding Model
__[Do it! Practice] Vector DB and Retriever
__[Do it! Practice] Generating Answers with a Language Model Based on Given Chunks
__[Do it! Practice] Implementing Query Expansion
_09-3 Completing a Chatbot with Streamlet
__[Do it! Practice] Adding a Retriever to Basic Streamlet Code
__[Do it! Practice] Citation

Chapter 10: Creating a Chatbot that Answers Questions Using Internet Search
_10- 1 Answer after searching the Internet - DuckDuckGo Search
__[Do it! Practice] Adding Internet Search Functionality to GPT
__[Do it! Practice] Setting Options for the Search Function
__[Do it! Practice] Get the article link
__[Do it! Practice] Importing Only Specific Areas Using Beautiful Soup
_10-2 Writing an article after researching materials - Tabili Search
__[Do it! Practice] Using Tabili
__[Do it! Practice] Become a journalist who researches and writes articles online.
_10-3 Summarizing YouTube videos
__[Do it! Practice] Searching YouTube with the YoutubeSearch package
__[Do it! Practice] Importing YouTube Subtitles with the YoutubeLoader Package
__[Do it! Practice] Summarizing the subtitle content
_10-4 Creating a Chatbot Using Web and YouTube Search
__[Do it! Practice] Adding a Web Search Tool to Your Chatbot
__[Do it! Practice] Adding a YouTube Search Tool

Chapter 11: Using the DeepSearch-R1 Model Locally
_11-1 Learn about the Deep Seek model
__The emergence of small-scale language models
__Deepseek-R1 model
__[Do it! Practice] Installing the Olama and DeepSeek-R1 Models
_11-2 Using the DeepSearch Model in LangChain
__[Do it! Practice] Building a Chatbot with DeepSearch and LangChain
_11-3 Creating a RAG based on DeepSeak
__[Do it! Practice] Creating a RAG with DeepSecure

Fourth Yard | Building a Collaborative AI Team Using Langgraph

Chapter 12: Getting to Know Langgraph
_12-1 Basic Chatbot Created with Langgraph
__What is a langgraph?
__Basic concepts of Langgraph - nodes, edges, and states
__[Do it! Practice] Creating a Simple Chatbot with LangGraph
__[Do it! Practice] Defining a State
__[Do it! Practice] Creating a Node
__[Do it! Practice] Setting up Edge
__[Do it! Practice] Outputting a Stream
_12-2 Memory for storing conversation content
__[Do it! Practice] Utilizing LangGraph's Memory Function
_12-3 Creating a chatbot that writes articles after searching the Internet
__[Do it! Practice] Creating a Newspaper Reporter Chatbot
__[Do it! Practice] Setting up a router
__[Do it! Practice] Testing Tools and Writing Articles

Chapter 13: Building a Multi-Agent RAG Using Langgraph
_13-1 Preliminary work for RAG based on lang graph
__Multi-agent system and accurate guide
Improving the limitations of __RAG
__[Do it! Practice] Preprocessing PDFs and Creating a Vector Database
_13-2 Getting to know the router
__Router
__[Do it! Practice] Setting up a router in a chatbot
_13-3 Creating a RAG Agent with Langgraph
__[Do it! Practice] Filtering Only Relevant Chunks
__[Do it! Practice] Creating RAG Answers
_13-4 Defining a graph
__[Do it! Practice] Declaring Graph States and Defining Nodes
__[Do it! Practice] Creating a StateGraph
__[Do it! Practice] Multi-Agent Testing

Chapter 14: Creating a Multi-Agent that Generates a Table of Contents with LangGraph
_14-1 Agent creating a table of contents with users
__Multi-agent created in this chapter
__Communicator agent that communicates with users
__[Do it! Practice] Creating a Communicator Agent
__Content strategist agent who writes the table of contents for a book
__[Do it! Practice] Create a content strategist agent, content_strategist, to create a table of contents.
_14-2 Supervisory Agent Acting as a Facilitator
__We need a boost! ― Supervisor Agent
__[Do it! Practice] Adding a Supervisor Agent
__[Do it! Practice] Defining Output Formats with Pydantic's BaseModel
_14-3 Vector search agent utilizing web search and RAG
__Using Vector DB to Efficiently Search the Web
__[Do it! Practice] Creating a Web Search Function
__[Do it! Practice] Creating a Vector DB
Vector search agent that finds high-relevance chunks
__[Do it! Practice] Connecting to LangGraph
_14-4 Web search agent for missing information
__A web search agent that finds missing information
__[Do it! Practice] Creating a web search agent, web_search_agent
__[Do it! Practice] Using Search Results in the Table of Contents

Chapter 15: Building a Multi-Agent That Makes Its Own Decisions and Works
_15-1 Creating a common goal for agents
__Business analyst agent checking goals
__[Do it! Practice] Creating a business_analysist agent that understands user intent
_15-2 Establishing clearer guidelines with templates
__Template that defines the document format and derives the response format
__[Do it! Practice] Creating a Template for Creating a Table of Contents
__[Do it! Practice] Developing System Prompts Using a Table of Contents Template
__Multi-agent that judges and works on its own
__[Do it! Practice] Building a Multi-Agent System That Makes Its Own Decisions and Works
_15-3 Developing an agent who reviews and edits on his own
__Table of Contents Review Agent
__[Do it! Practice] Adding advice items to the table of contents and reflecting them in business_analyst
__[Do it! Practice] Creating an outline_reviewer to review the table of contents
__[Do it! Practice] Configuring the Vector Search Agent to Advise the Business Analyst Agent
__[Do it! Practice] Preventing an Infinite Loop

Chapter 16: Using Artificial Intelligence More Safely
_16-1 Running the Llama and Embedding Models Locally
__[Do it! Practice] Running Meta's Llama Model Locally
__[Do it! Practice] Creating a Simple Llama-Based Chatbot
__[Do it! Practice] Using Local Embedding Models
_16-2 Developing LLM-based services
__Quick answers vs. answers the user wants
__Once the function development is finished, that's when it starts.

Search

Using GPT API, DeepSearch, Lama, LangChain, LangGraph, and RAG
Create an AI agent that changes your daily life with your own hands!

The use of LLM and generative AI technologies is rapidly becoming standardized.
This book systematically introduces cutting-edge methodologies and builds a solid foundation for applying AI technology, while leveraging LLM to create AI agents.
In the first yard, we will learn what LLM is and set up a Python environment to develop AI agents.
In the second yard, we'll use the GPT API to create programs that automate various tasks, such as summarizing papers, writing meeting minutes, and analyzing stocks.
In the third yard, we will implement multi-agent using LangChain and RAG and use the DeepSearch language model locally.
In the fourth yard, you will learn how to create a multi-agent that uses LangGraph to create a table of contents and how to utilize the Llama language model and embedding model locally without worrying about security.
By following the practical training, you will not only understand the current technology, but also lay the foundation for easily accepting new technologies that will continue to emerge in the future.


The era of one AI agent per person!
Create everything from custom business automation programs to multi-agent solutions!

This book explores how to create custom, time-saving automation programs using the GPT API, and how to develop multi-agents that utilize Langchain and Langgraph to make decisions and improve tasks. By understanding the principles of AI agent development, you can create your own custom AI agent.

***6 AI Agents Created in This Book***

1. AI researchers preprocessing and summarizing PDF documents
2.
AI secretary organizing meeting minutes
3.
AI image analyst that analyzes images and creates problems
4.
AI Investor that provides the latest stock information
5.
A personalized AI agent that searches documents, YouTube, and the internet to answer your questions.
6.
AI agents that create tables of contents for books and reports

If you know the basics of Python, let's get started!
Learn practical AI development with friendly explanations and problem-solving methods!

This book is designed to help even those who are new to AI technology learn how to develop AI agents and effectively utilize them in their daily lives.

1.
We explain the practical process and code in detail so that even beginners can easily follow along.

2.
We provide result files so you can learn by comparing them with your own code.
3.
In the "One Step Further!" corner, you can learn not only confusing concepts but also tips to improve your skills.
4.
It goes beyond simply providing a manual, and includes all the problems that may arise during the actual development process and the process of resolving them.

I recommend this book to these people!

ㆍ Those who want to apply AI agents using LLM to practical work
ㆍ Those who want to systematically understand the latest AI technologies such as LLM and generative AI
Developers and planners who need to design and develop AI services

Set a goal and study systematically on your own!
30-day and 15-day courses, two study plans and example files provided!

This book provides a study plan so that you can study systematically on your own.
Write your own target dates on your study plan and start studying. If you're new to LLM development, we recommend the 30-day course. If you want to quickly master AI agent development, we recommend the 15-day course.
The 15-day study plan can cover a semester's worth of classes! Sample files for this book are also available on the Aegis Publishing website and the author's GitHub account.
You can study without any obstacles by comparing the code you wrote with the example files you downloaded!

Come to the 'Do it! Study Room' where we learn, share, and grow together!

If you plan and study alone, you will quickly become tired.
Share difficult content with readers who have similar concerns.
It's more rewarding to study while helping each other.
By helping others with the things I understand well and receiving help with the things I don't understand well, I can enjoy the effect of reviewing while studying.
If you want to go far, go with a friend.

Do it! Study Room: cafe.naver.com/doitstudyroom