N8n
Fair-code workflow automation with native AI — a visual canvas plus code, 400+ integrations, and LangChain-based agent nodes; self-host free or cloud per-execution.
n8n (~192k stars) is the automation platform that grew an AI brain: a visual workflow canvas (with code when you want it), 400+ app integrations, and AI agent nodes — built on LangChain — with memory backends, vector-store nodes for RAG, and broad model support. Fair-code licensed: free self-hosting for internal use, EUR-priced cloud billed per execution.
n8n attacks AI from the opposite direction of the AI-native platforms: it was already the automation layer — ~192k stars, 400+ integrations, a decade of workflow muscle — and then gave its workflows a brain. The result is distinctive: agents with hands, where the AI node sits between real triggers and real actions.
Highlights
- AI Agent nodes on LangChain — Tools, Conversational, ReAct, Plan-and-Execute, and SQL agent types as canvas nodes, with chains for Q&A and summarization.
- The integration moat as a toolset — 400+ apps and 900+ templates double as agent tools: the agent that reads the ticket, queries the DB, and posts to Slack is three nodes.
- RAG without leaving the canvas — vector-store nodes (Pinecone, Qdrant, Chroma, Weaviate), document loaders, embeddings, retrievers.
- Memory backends — Simple to Redis/Postgres/MongoDB/Zep for stateful conversations.
- Visual + code — the canvas for structure, Code nodes for the parts that are genuinely code (now sandboxed in 2.0's task runners).
- Self-host or cloud —
npx n8n/ Docker for free internal self-hosting; cloud priced per execution (unlimited steps and users), EUR-denominated.
In an AI-assisted workflow
docker run -it --rm -p 5678:5678 -v n8n_data:/home/node/.n8n docker.n8n.io/n8nio/n8n
# editor at localhost:5678 — drop an AI Agent node into any workflowThe signature pattern: automation-first, intelligence where it pays — a deterministic pipeline with one judgment step (classify, draft, decide) handled by an agent node, with human approval nodes guarding the consequential branches.
NOTE
License clarity: free for internal business use (including commercial), self-hosted; reselling n8n — hosting it for customers, embedding in paid products — needs a license. The .ee. files in the repo are enterprise-licensed despite being public.
Good to know
The October 2025 Series C ($180M, Accel-led, $2.5B valuation, NVIDIA's venture arm participating) made n8n the automation category's AI flagship; 2.0 followed with security-by-default. Cloud "AI credits" are starter allowances — you bring model keys. The head-to-head with the AI-native canvas: n8n vs Dify.
Frequently asked questions
- What makes n8n good for AI workflows specifically?
- It puts agents inside real automations. The AI Agent node (Tools, ReAct, Plan-and-Execute, SQL variants) is built on LangChain, with conversation memory (Redis/Postgres/Zep), vector-store nodes (Pinecone, Qdrant, Weaviate…) for RAG, and every major model provider — and crucially, 400+ app integrations as the agent's hands: the AI step slots between the Gmail trigger and the Slack action.
- Is n8n open source?
- Fair-code, not OSI open source. The Sustainable Use License allows free use and modification for internal business purposes (commercial included) and personal use — but you can't sell n8n hosting or embed it in a paid product without an enterprise/embed license, and .ee.-flagged files in the public repo require enterprise licensing.
- What changed in n8n 2.0?
- A security-hardening major (December 2025): Code nodes run in isolated task runners and lose env-var access by default, arbitrary-command nodes ship disabled, plus a Publish-vs-Save deployment model and large performance work. It's breaking for older self-hosted setups — run the migration report before upgrading.
Related
- n8n vs Dify: Which AI Workflow Platform? (2026)Automation-first vs AI-native — n8n's 400+ integrations with agent nodes vs Dify's LLM-app platform with built-in RAG. Licenses, pricing shapes, and the fit test.
- DifyThe visual platform for LLM apps and agentic workflows — canvas-built chatflows, RAG pipeline, agent nodes with 50+ tools, and LLMOps, self-hosted via Docker.
- LangchainThe provider-agnostic agent framework, post-1.0: a standard create_agent loop on the LangGraph runtime, middleware hooks, and the largest integration ecosystem.
- Human-in-the-Loop (HITL)Human-in-the-loop design inserts human judgment at decisive points in an AI workflow — approving actions, resolving ambiguity, owning the irreversible steps.
- Which Agent Framework in 2026? LangGraph vs CrewAI vs AutoGen vs OpenAI Agents SDK vs Claude Agent SDKA decision guide to the major AI agent frameworks — control vs. abstraction, multi-agent models, state and durability, and which fits your project.