n8n self-hosted: the open-source Zapier alternative on your VPS
n8n is the open-source, self-hosted Zapier alternative that developers and automation teams have been moving to in 2026. Same concept as Zapier or Make (formerly Integromat): connect services together, automate the boring stuff, trigger actions on events. The difference is that n8n is self-hosted on your VPS with no per-execution fees and no limits on workflow count.
This guide covers what makes n8n a good Zapier alternative, what you can build with it, what resources you need, and how to deploy it on a VPS in a few minutes.
What n8n does
n8n (pronounced “n-eight-n”) is a workflow automation platform with 400+ native integrations. If you’ve used Zapier or Make, it’s the same concept: when X happens in one tool, do Y in another. Visual editor, drag-and-drop nodes, branching logic, error handling, everything you’d expect from a modern automation tool.
The difference from Zapier is that n8n is open source, self-hosted, and doesn’t charge per execution. Build as many workflows as you want, run them as often as you need, no pricing tiers.
Why n8n self-hosted is a better Zapier alternative
n8n offers a cloud version, but self-hosting gives you:
- Unlimited executions. Zapier and n8n Cloud both charge based on how many times your workflows run. Self-hosted n8n has no limits.
- Your data on your server. Workflow data, credentials, and execution logs stay on your VPS.
- No per-workflow costs. Build hundreds of workflows without the bill going up.
- Custom nodes and modifications. Install community nodes, write your own integrations, or modify n8n itself.
- Better performance for AI workloads. Keep AI model API calls close to your other infrastructure.
Trade-off: you’re responsible for updates, backups, and keeping the server running. On a VPS with Docker, that’s minimal work.
Why n8n has taken off for AI automation
n8n has always been good at connecting APIs. In the last year, it’s become one of the go-to platforms for AI automation. The AI nodes let you:
- Chain LLM calls into multi-step reasoning workflows
- Connect to vector databases for RAG (retrieval-augmented generation)
- Build AI agents that decide which tools to use based on the task
- Process documents by extracting, summarizing, and classifying content with AI
You can build workflows like “When a customer emails support, classify the intent with Claude, look up their order in the database, draft a response, and send it for human review” without writing code.
What people build with n8n as a Zapier alternative
Business automation
- Sync data between CRM, accounting, and project management tools
- Auto-generate invoices from completed tasks
- Monitor website uptime and send alerts
- Process form submissions and route them to the right team
AI workflows
- Classify incoming emails and route to departments
- Summarize meeting transcripts and post to Slack
- Extract data from PDFs and populate spreadsheets
- Build internal chatbots powered by company documentation
DevOps
- Monitor server health and trigger alerts
- Auto-deploy when code is pushed to main
- Sync GitHub issues with project boards
- Generate changelog entries from commits
Content and marketing
- Auto-post to social media on a schedule
- Generate content briefs from keyword research
- Monitor brand mentions and compile reports
- Resize images and optimize for different platforms
VPS requirements for n8n self-hosted
n8n is lightweight compared to most self-hosted platforms:
| Workload | RAM | vCPUs |
|---|---|---|
| Light use (few workflows) | 2 GB | 1 |
| Medium use (dozens of workflows) | 4 GB | 2 |
| Heavy use (AI workflows, high frequency) | 8 GB+ | 4+ |
A basic VPS handles n8n comfortably. You only need more resources for lots of AI-heavy workflows or large data volumes.
Quick n8n self-hosted setup
The Docker method is simplest:
- Deploy an Ubuntu VPS with at least 2 GB RAM
- Install Docker
- Run n8n with a single command:
docker run -d --restart unless-stopped \
-p 5678:5678 \
-v n8n_data:/home/node/.n8n \
n8nio/n8n- Access the web editor at
http://your-vps-ip:5678 - Create your admin account
That’s it for a basic setup. For production, add a reverse proxy with SSL (Nginx + Certbot), use PostgreSQL instead of the default SQLite, and set up regular backups.
n8n vs Dify vs Zapier
Quick comparison:
- Zapier/Make: SaaS, easy, but pay-per-execution gets expensive fast. Good if you don’t want to manage a server.
- n8n: Self-hosted, 400+ integrations, AI nodes. Best general-purpose Zapier alternative with some AI mixed in.
- Dify: Self-hosted, focused specifically on AI applications (RAG, chatbots, AI agents). Best for building AI-first products.
Some teams run both n8n and Dify: n8n for integration and automation, Dify for AI-heavy applications.
The n8n + Ollama combo
One of the most popular self-hosted stacks right now is n8n connected to Ollama on the same VPS. Ollama gives you local AI inference; n8n gives you the workflow engine. Together, AI-powered automation that runs entirely on your hardware with no API costs.
Requires more RAM (8-16 GB to run both n8n and a decent model), but cost savings over API calls can be significant for high-volume processing.
Deploy a Linux VPS and have n8n self-hosted running in 5 minutes. For AI-heavy workflows, pair it with a larger VPS in any of our six locations.