Self-hosted Zapier alternative: unmetered automation with n8n

Automation is the quiet infrastructure of a working team -- the thing that moves a lead into your CRM, posts the alert, files the invoice, syncs the two tools that refuse to talk. Build enough of it in Zapier or Make and you hit the catch: every workflow lives on their servers, and every time one fires it spends against a task or operation quota. The moment your automations get genuinely useful, they get expensive. And the credentials they carry -- API keys, database logins, OAuth tokens to half your stack -- sit in someone else's cloud. n8n is the version you own.

What n8n is

n8n is a full workflow automation platform you run yourself. You build in a visual editor -- drag nodes onto a canvas, wire a trigger to a chain of actions, watch real data flow through each step -- and it ships with hundreds of pre-built integrations for the SaaS tools you already use. When there isn't a dedicated node, a generic HTTP Request node talks to any API you can reach, and a Code node lets you drop into JavaScript or Python for the parts that need real logic.

It is fair-code and source-available, self-hostable for free, and it is not a stripped-down clone. This is the same product powering production automations at companies that pay for the cloud tier; you're just running it on your own machine. Compare that to Zapier's free plan capping you at a hundred tasks a month and two-step Zaps, or Make charging per operation -- n8n has no execution meter at all when you host it.

The features that actually matter

• The visual node editor. Drag nodes, connect them, and see the exact data passing between each step. You build and debug in the same view, no separate test runs to burn quota on.
• Hundreds of integrations. Pre-built nodes for Slack, Google, Notion, Airtable, HubSpot, Stripe, GitHub, Postgres and hundreds more -- the app directory Zapier sells access to, without the tier gates.
• Webhook and schedule triggers. Kick off a workflow from an inbound HTTP call, or on a cron schedule you define. Instant webhooks are a premium feature elsewhere; here they're just a node.
• The HTTP Request node. Call any REST or GraphQL API even when no dedicated integration exists. This is the escape hatch that turns "unsupported app" into "supported."
• The Code node. Write inline JavaScript or Python for transforms, loops, and logic a point-and-click builder can't express. Zapier's code steps sit behind a paid plan.
• AI and LangChain nodes. Build AI agents, chain LLM calls, and wire up retrieval and vector stores natively -- automation and AI on one canvas rather than a bolt-on.
• Encrypted credentials. Store an API key or OAuth grant once, reference it across every workflow, encrypted at rest with your own key -- never re-pasted into a form.
• Real control flow. Sub-workflows, merges, wait steps, and dedicated error workflows that catch a failed run and route it somewhere useful, plus full execution history showing input and output for every node.

Why self-host it instead of the hosted version

Your data and credentials stay yours. Every automation you build touches something sensitive -- a customer record, an internal API key, a payment webhook. Self-hosted, all of it lives on your server and passes through your infrastructure, never a third party's execution engine.

Cost that doesn't scale per run. This is the big one. Zapier meters tasks and Make meters operations, so a busy workflow is a growing bill. On your own box, an execution is free whether you run a thousand a month or a million -- you pay for a server, not for the work it does.

Control and branding. Set your own environment variables, point it at your own domain, install community nodes the cloud tier restricts, and switch on queue mode with workers when you need to scale. It's your instance, tuned to your workload, not a seat in someone else's tenancy.

Who it's for

Agencies and ops teams running dozens of automations for clients or internal processes, where a per-task bill becomes the single largest line item long before the work justifies it.

Developers who want the Code node and the HTTP Request node -- people who'd rather write ten lines of JavaScript than fight a rigid builder, and who need to hit APIs no SaaS connector supports.

Privacy-sensitive organisations -- anyone moving customer data, health records, or internal secrets through their workflows who cannot hand those credentials to a third-party cloud in the first place.

The honest setup notes

Self-hosting n8n well means owning a few real chores. It defaults to a SQLite database, which is fine to kick the tyres but wants PostgreSQL for anything you depend on -- so that's a database to run and back up. There's an encryption key that protects all your stored credentials; lose it and every saved connection is gone, so it belongs in your backups from day one. Webhooks need a stable public HTTPS URL and a reverse proxy in front, or half your triggers never fire. n8n ships updates often, so you're on the hook for keeping the image current. And if you want password resets and user invites to work, that's SMTP to configure too.

None of it is hard. All of it is yours to remember, monitor, and fix at the wrong hour. Or let us run it.

Deploy it in one click

Rolling your own means a Docker host, a Postgres instance, a reverse proxy with a real certificate, the webhook URL wired correctly, the encryption key stashed safely, and a plan for updates -- an afternoon if it goes smoothly, longer if it doesn't. On Caliber Node you deploy n8n in one click and skip every step of that. We provision the container, attach a real database, terminate TLS on your domain, set the webhook URL, and keep the thing patched. You get a working n8n at your own address with encrypted credentials, execution logs, and SFTP into the files -- and none of the plumbing to babysit. It's the full self-hosted platform with the operational tax paid for you.

First run

1. Open your n8n instance and create the owner account -- this is your admin login, held on your server, not a shared cloud tenant. Do it straight away: until you have, anyone who finds the address could claim it.
2. Create a new workflow and add a trigger node: a Schedule trigger to run on a timer, or a Webhook trigger to fire on an inbound HTTP call.
3. Add an action node -- pick one of the built-in integrations, or drop in an HTTP Request node to hit any API you like.
4. Open the connected app's node, add its credential once, and reference it. It's stored encrypted from here on.
5. Run the workflow to execute it live, watch the data move through each node, then toggle it active. Your first automation is now running on hardware you control.

The bigger picture

Automation quietly became the layer that holds a business together, and the SaaS model turned that layer into a subscription that grows every time you lean on it. Owning your automation platform flips it back: the workflows are yours, the credentials are yours, the executions cost nothing beyond the server they run on, and nobody can raise the price of work you've already built. n8n is one of the clearest cases where self-hosting wins outright -- you keep every bit of the capability and gain the whole thing to own.

Deploy n8n in a few minutes →

Or browse the full app catalog to see what else you can own.

One-click self-hosted apps, managed for you -- real logs, SFTP, and your data always exportable. Never a black box.