In an era where Large Language Models (LLMs) are becoming increasingly widespread, the ecosystem of tools surrounding them is also rapidly evolving. While LLMs are capable of impressive automation, they don’t always make fully accurate decisions—especially when it comes to high-stakes tasks or those requiring a high degree of reliability. Human-in-the-Loop (HITL) is a technique that allows humans to intervene in an AI system’s processing pipeline—confirming, editing, or providing additional information when necessary. With LangGraph, you can use the interrupt() function to pause execution at predefined checkpoints and wait for human feedback before proceeding. Once input like “accept” or “edit” is received, the process can resume using Command(resume=...), transitioning the state forward. This technique is already being applied in systems like Claude Desktop and Cursor Editor, where it helps ensure that all critical actions and content are subject to human oversight. In this article, I’ll walk you through how to implement Human-in-the-Loop using LangGraph (for logic flow and checkpoints) and Chainlit (to build a manual interaction interface). The goal is to help you establish a clear, moderated AI workflow—enabling efficient human control and feedback.

Nowadays, AI is widely applied in various fields, especially in daily tasks for office workers and programmers. While AI excels at analyzing and understanding data, it still faces limitations in storing information, retrieving external data, and interacting with the surrounding environment.

In this article, I will introduce a method that helps AI models expand their data access capabilities and interact more flexibly with system entities.

Nowadays, AI has become an indispensable tool in daily work, especially for office workers and developers. However, current AI models still face significant limitations in accessing and processing information.

Therefore, in this article, I will introduce a method to enhance AI’s information retrieval capabilities and improve the efficiency of its data access.

Python is a versatile language programming, easy to learn because of clean syntax. However, Python has also some secrets that not everyone know. In this post, I show you what are secrets of Python?

Inversion of Control and Dependency Injection are two important concepts in object-oriented programming. It is the basic principles that help us build flexible, maintainable, and extensible applications. This article will help you understand more about Inversion of Control and Dependency Injection.

What are the differences between threads in Python and threads in other programming languages? What is the GIL? Is it always more efficient to use multithreading and multiprocessing over single-threading and single-processing? In this article, I will discuss the limitations of threads in Python and provide recommendations on when to use multithreading, multiprocessing, or stick with single-threading.

Design patterns are among the essential skills for a programmer. Mastering the application of design patterns not only enhances their understanding of programming but also elevates their value in the industry. In this article, I'll share one of the relatively common design patterns and how to apply it in frontend web development - the Observer pattern.

Since ChatGPT has become increasingly popular, chatbot is appearence everywhere in your life. In this post, I want to show you how to create a chatbot in locally with Python.

Asynchrony can be implemented in many programming languages, including Python. However, asynchrony in early Python was not as widely used as Javascript or C# because of its difficult implementation. In this article, I will introduce the history of asynchronous programming in Python and its growth at the present time.

Recently, there has been a web framework become popular in the Python community - FastAPI. In this post, I show you how FastAPI is good and why you should decide on FastAPI for development APIs