The Current State of AI Development: What are "LangChain / LangGraph"? A Quick Explanation from a Historical Perspective
Recently, you may have heard the terms "LangChain"and"LangGraph"in the AI world. In short, these are"mechanisms for making AIs like ChatGPT operate more intelligently and complexly."
I have summarized the historical lineage (flow) from which these emerged from three perspectives, written so that anyone can understand them without specialized knowledge.
1. Lineage from the perspective of AI evolution: From one-way "chains" to "teams" that think for themselves
The way we use AI has evolved dramatically over the past few years.
1st Generation: Simple Chat (2022–) A simple Q&A format where humans ask questions and AI answers.
2nd Generation: Serial "Chains" (2023–) 👈 LangChain It became possible to automate AI tasks by connecting them like a single chain, such as "search, summarize, and translate."
3rd Generation: Complex "Webs" (2024–Present) 👈 LangGraph It has evolved into complex interactions (AI agents) like a web, where the AI can "try again if the result is bad" or "discuss among multiple AIs."
AI, which was once just a "tool," has evolved into a lineage that acts like an "autonomous subordinate."
2. Lineage from the perspective of IT history: A "toolbox" that makes things accessible to everyone
In the world of IT, the same history always repeats itself whenever a new technology emerges.
Early stage: Experts write difficult spells (code) on black screens.
Adoption stage: "Convenient toolboxes (frameworks)" appear that make it easy for anyone to use.
LangChain and LangGraph are precisely part of this "lineage of toolboxes that appeared during the adoption stage."
Thanks to their emergence, engineers around the world can now build "convenient smartphone apps and internal systems using AI" at an overwhelming speed.
3. Lineage from the perspective of manufacturing philosophy: From "conveyor belts" to "conference rooms"
It is easy to understand the philosophy of these mechanisms by comparing them to factory lines.
LangChain is the lineage of the "conveyor belt" It is a system where data flowing from upstream is processed by AI in sequence and sent downstream. It is a one-way street, so you don't get lost, but you cannot go back if a problem occurs along the way.
LangGraph is the lineage of the "conference room" Members (multiple AIs) discuss, send tasks back to previous steps if they are not good enough, and proceed to the next step if they are OK. It has realized such "complex workflows that go back and forth."
Summary: What will change?
Until now, AI was an "excellent assistant." However, with the emergence of the LangChain and LangGraph lineage, "AI teams that can organize their own work and run until completion" will be increasingly integrated into business settings from now on.
Behind the scenes of technology, a 'revolution in mechanisms' is currently
taking place.
#AI #GenerativeAI #ChatGPT #LangChain #LangGraph #AIAgent #Technology #Business #HowToWriteNote
いいなと思ったら応援しよう！
ここまで読んでいただけるとは✨チップはnote創作に使わせていただきます🤗