构建工作流
What chains do
Chains1 bring together different components of AI to create a cohesive system. They set up a sequence of calls between the components. These components can include models and memory2 (though note that in n8n chains can't use memory).
Chains in n8n
n8n provides three chain nodes:
- Basic LLM Chain: use to interact with an LLM, without any additional components.
- Question and Answer Chain: can connect to a vector store3 using a retriever, or to an n8n workflow using the Workflow Retriever node. Use this if you want to create a workflow that supports asking questions about specific documents.
- Summarization Chain: takes an input and returns a summary.
There's an important difference between chains in n8n and in other tools such as LangChain: none of the chain nodes support memory. This means they can't remember previous user queries. If you use LangChain to code an AI application, you can give your application memory. In n8n, if you need your workflow to support memory, use an agent. This is essential if you want users to be able to have a natural ongoing conversation with your app.
Footnotes
- Back
AI chains allow you to interact with large language models (LLMs) and other resources in sequences of calls to components. AI chains in n8n don't use persistent memory, so you can't use them to reference previous context (use AI agents for this).
- Back
In an AI context, memory allows AI tools to persist message context across interactions. This allows you to have a continuing conversations with AI agents, for example, without submitting ongoing context with each message. In n8n, AI agent nodes can use memory, but AI chains can't.
- Back
A vector store, or vector database, stores mathematical representations of information. Use with embeddings and retrievers to create a database that your AI can access when answering questions.
官方原文和授权
本页来自 N8N 英文官方网站固定快照，并转换成 xueai 静态页面。内容以 N8N 持续更新的官方页面为准。