n8n
Self-hosted workflow automation with native LLM, vector DB, and agent nodes - the local Zapier for AI pipelines.
What it does
Core capabilities at a glance
- 500+ pre-built integrations
- Native Ollama, OpenAI, Anthropic, vector DB nodes
- LangChain-style agent nodes
- Visual workflow builder
- Self-hostable on a Pi
- Cron + webhook triggers
- Code nodes for JS/Python
- Built-in queue + retries
Deep dive
The full breakdown - performance, comparisons, and setup
n8n
Short answer: n8n is the local-first answer to Zapier/Make, and as of 2024 it grew first-class AI nodes (LangChain agents, vector DBs, Ollama, embeddings) that make it the cleanest way to build production AI workflows without writing glue code.
What it is
n8n (pronounced "n-eight-n", short for "nodemation") is a node-based workflow automation platform. Drag triggers (webhook, cron, email, file watch), connect them to action nodes (HTTP, database, Ollama, Slack, 500+ others), and you've built a workflow that runs on your own infrastructure.
Source-available under the Sustainable Use License - free for self-hosting, commercial restrictions only if you're reselling n8n itself as a service.
Why local-AI builders use it
- AI nodes are first-class - not bolted on. Native Ollama, OpenAI-compatible, embeddings, Qdrant, Pinecone, agent loops
- Visual debugging - see exact JSON at every node, replay from any step
- Self-hosted means private - no data leaves your network; runs on a Pi if you want
- Code escape hatch - drop into JS or Python whenever the visual nodes aren't enough
- Free, forever - no per-task pricing like Zapier
Real-world AI workflows people build
- Personal RAG bot: webhook → embed query → Qdrant search → Ollama → response
- Email triage: IMAP trigger → classify with LLM → file/respond/snooze automatically
- Content monitoring: RSS → summarize with local LLM → Discord post
- Lead scoring: form webhook → enrich → LLM scoring → CRM upsert
- Multi-agent research: cron → research agent → write agent → review agent → publish
Hardware sizing
n8n itself is light. Almost all load is on your LLM backend.
| Setup | Suitable for |
|---|---|
| Pi 5 / 8 GB mini-PC + cloud LLM | Light hobby automation |
| 16 GB mini-PC + Ollama + 7-8B model | Solo dev, ~100 runs/day |
| RTX 4090 + Ollama + Qwen3 30B | Small team, ~1k runs/day |
| Dual GPU server + vLLM | Production team, 10k+ runs/day |
How it compares
| n8n | Zapier | Make (Integromat) | Activepieces | Langflow | |
|---|---|---|---|---|---|
| Self-hostable | ✓ | ✗ | ✗ | ✓ | ✓ |
| AI agent nodes | ✓✓ | basic | basic | basic | ✓✓ (agent-focused) |
| Pricing | free self-host | $20-600/mo | $9-30/mo | free self-host | free self-host |
| Best for | AI pipelines + general automation | non-technical, SaaS-only | complex branching | self-host alt to Zapier | agent-only flows |
Frequently asked
Is n8n really free?
For self-hosting, yes - Sustainable Use License. Cloud-hosted (managed) is paid. We assume self-host throughout this site.
How does it compare to Langflow / Flowise for AI agents?
Langflow and Flowise are agent-first; n8n is automation-first with strong agent support. If your workflow is 80% agent-loop and 20% glue, use Langflow. If it's 30% AI and 70% other systems, n8n.
Can it call my local Ollama?
Yes - there's a native Ollama node. Or use the HTTP node and point at http://localhost:11434/api/generate.
Get started
docker run -d --name n8n \
-p 5678:5678 \
-v ~/.n8n:/home/node/.n8n \
docker.n8n.io/n8nio/n8nVisit http://localhost:5678. Create your account, drop a webhook node, connect it to an Ollama node, click "test workflow". You now have a local AI HTTP API.
Frequently asked
Quick answers to common questions
What is n8n?
n8n is a automation tool for local AI workloads. Self-hosted workflow automation with native LLM, vector DB, and agent nodes - the local Zapier for AI pipelines.
Is n8n free and open source?
Yes, n8n has 191,487 GitHub stars and is licensed under Sustainable Use (source-available, free for self-host). You can self-host it for free on docker, linux, macos, windows.
What platforms does n8n support?
n8n runs on docker, linux, macos, windows.
What hardware do I need for n8n?
The hardware requirements depend on which models you run. Check our hardware directory for compatible GPUs and systems. n8n has 191,487 GitHub stars and an active community.
Does n8n support GPU acceleration?
n8n supports GPU acceleration via CUDA, Metal, or Vulkan depending on your platform. For the best performance, pair it with an NVIDIA RTX 4090 or 5090.
What are the best alternatives to n8n?
Popular alternatives include other automation tools in our directory. Browse our full collection at /tool for comparisons, community reviews, and benchmark data to find the right fit for your workflow.
How much does n8n cost?
n8n is free-open-source. It is completely free and open source to self-host.
Pairs well with
Complementary tools, models, and hardware
Similar tools
More tools like this one
Comments coming soon
Configure NEXT_PUBLIC_GISCUS_REPO_ID and NEXT_PUBLIC_GISCUS_CATEGORY_ID at giscus.app to enable.