LangChain
LangChain is an open-source Python and JavaScript framework that provides abstractions and components for building applications that use large language models, including chains, agents, memory, and tool integrations.
理解する LangChain
Released in October 2022, LangChain became one of the most widely adopted frameworks in the LLM application ecosystem. It introduced standardized abstractions for the most common patterns in LLM-powered applications: connecting a model to external data sources, chaining multiple prompts together, enabling LLMs to use tools, and persisting context across interactions. LangChain's core primitives include: **Chains**: Sequences of LLM calls and other operations composed into a pipeline. A chain might retrieve relevant documents, inject them into a prompt, call the LLM, and parse the output. **Agents**: LLM-driven decision loops that choose which tools to use and in what order to accomplish a goal. LangChain popularized the ReAct (Reasoning + Acting) agent pattern. **Memory**: Mechanisms for persisting context across multiple LLM calls, from simple conversation buffers to vector-store-based long-term memory. **Tool integrations**: A large ecosystem of pre-built connectors to external APIs, databases, and services. As agent use cases became more sophisticated, the LangChain team built LangGraph as a separate library for complex, stateful, multi-agent workflows. LangGraph provides more explicit control over agent execution flow using a graph-based model, addressing limitations of LangChain's sequential chain abstraction for production agentic systems. LangChain remains widely used for prototyping and simpler LLM applications, while LangGraph is preferred for production-grade agent systems that require fine-grained control, human-in-the-loop workflows, and persistent state.
GAIAの活用方法 LangChain
GAIA's backend is built on LangGraph rather than LangChain's chain-based abstraction, giving it fine-grained control over the multi-agent execution graph that orchestrates tasks across 50+ integrations. LangGraph's stateful, graph-based approach is better suited to GAIA's complex, long-running agent workflows than LangChain's simpler sequential chains. The LangChain ecosystem's tool integrations and community patterns informed GAIA's architecture.
関連概念
LangGraph
LangGraphは、サイクル、分岐、条件付きロジック、永続的な状態管理を備えた複雑なワークフローをサポートする、ステートフルなマルチエージェントAIアプリケーションを構築するためのフレームワークです。
Agent Loop
An agent loop is the iterative execution cycle of an AI agent in which it reasons about the current state, selects and executes an action (often a tool call), observes the result, and repeats until the task is complete or a stopping condition is reached.
Large Language Model (LLM)
A Large Language Model (LLM) is a deep learning model trained on massive text datasets that can understand, generate, and reason about human language across a wide range of tasks.
Function Calling
Function calling is a feature of AI models that allows them to generate structured, machine-readable invocations of predefined functions, enabling AI systems to reliably call external APIs and tools with the correct arguments.
AIオーケストレーション
AIオーケストレーションとは、単独では処理できない複雑なマルチステップタスクを完了するために、複数のAIエージェント、モデル、およびツールを連携させることです。
