n8n is a workflow automation platform built on a visual node graph. Each node is a trigger, an app integration, a data transform or a piece of control flow, and you wire them into a workflow that runs on a schedule, a webhook, a chat message or a manual trigger. It ships several hundred built-in integrations plus generic HTTP Request and Code nodes running JavaScript or Python, so anything without a dedicated node can still be reached by API — which is why it lands with technical users rather than pure no-code buyers.
Its AI layer is built on LangChain and uses a cluster-node model: a root node (AI Agent, Basic LLM Chain, Q&A Chain, Summarization Chain or a vector store) with sub-nodes attached for chat models, memory, tools, retrievers, embeddings, document loaders and output parsers. The AI Agent node picks which attached tools to call at runtime, loops through tool calls and evaluates results before answering. Any n8n workflow can itself be exposed as a tool, and MCP client and server nodes are supported.
It runs either as n8n Cloud or self-hosted on your own infrastructure, which is the main reason it gets chosen in regulated or data-sensitive environments. Typical buyers are internal ops and automation engineers, technical agencies, and dev teams building AI agents and internal integrations.
The pricing model is the point. An execution is a single run of your entire workflow — n8n's own FAQ states it does not matter how many steps are in the workflow or how much data it processes. A 40-step workflow costs the same as a one-step workflow, which is the opposite of Zapier's and Make's per-task billing. Only production runs count: manual runs, sub-workflow calls and error-workflow runs do not.
