n8n

What it does

n8n is a workflow automation platform aimed squarely at technical teams: it combines a visual, drag-and-drop canvas with the ability to write real JavaScript or Python inside any step, so teams aren’t stuck once a workflow needs logic a pure no-code tool can’t express. Its biggest structural difference from Zapier and Make is that n8n is source-available and can be fully self-hosted via Docker, which matters for teams that want workflow data to never leave infrastructure they control, or that want to avoid vendor lock-in on a business-critical process.

n8n has invested heavily in native AI agent building rather than treating AI as an add-on feature. It supports multi-agent setups, retrieval-augmented generation (RAG) pipelines, connections to multiple LLM providers, and the Model Context Protocol (MCP) for interoperating with AI coding tools and assistants — all visible and editable on the same canvas used for traditional automation. This makes it a common choice for teams that want to see and control every step of an agent’s reasoning rather than triggering an opaque black-box agent.

The tradeoffs are a steeper learning curve than Zapier or Make, and the operational overhead of managing your own server if you choose to self-host. n8n Cloud, the fully managed option, discontinued its permanent free plan in 2026, so anyone not self-hosting should expect to pay from the Starter tier up. The platform’s license (a “fair-code” Sustainable Use License) also isn’t fully permissive open source — it allows free self-hosting and internal modification but restricts some commercial redistribution use cases, which is worth understanding before building a product on top of it.

Ideal users

• Developers and technical teams who want full source-level control and the option to self-host rather than depend entirely on a vendor's cloud
• Teams that need custom code steps (JavaScript or Python) inside a workflow for logic a visual builder can't express
• Organizations building AI agents (RAG pipelines, multi-agent setups, MCP-based tool use) that want to see and control every reasoning step, not just trigger the agent as a black box

Who should avoid it

• You want the simplest possible no-code experience with zero technical setup — Zapier or Make have gentler learning curves for non-technical users
• You don't have anyone on the team willing to manage self-hosting infrastructure, and you also want to avoid n8n Cloud's per-execution pricing
• You need a mature, permissive open-source license for redistribution — n8n uses a fair-code/Sustainable Use License with restrictions, not a fully permissive open-source license
• You want a polished, fully managed AI-agent assistant experience out of the box — Lindy is closer to that than n8n's more hands-on canvas

Key features

• Visual workflow canvas with the option to write custom JavaScript or Python code inside any step
• 500+ pre-built integrations plus generic HTTP request and webhook nodes for anything else
• Free, unlimited self-hosted Community Edition via Docker, with full source code on GitHub
• Native AI agent building: multi-agent setups, RAG pipelines, multiple model support, and Model Context Protocol (MCP) support
• Developer-oriented debugging: step re-runs without replaying the whole workflow, data mocking, and native execution logs
• Unlimited active workflows on every plan (cloud and self-hosted), with cloud pricing based on execution volume instead

Pricing

Free — €20/month (Starter plan, billed annually) for n8n Cloud, or free to self-host (Community Edition, pay only for your own server)

n8n Cloud plans (2026), confirmed on n8n.io/pricing: Starter at €20/month (billed annually) for 2,500 executions, Pro at €50/month for 10,000 executions, and Business at €667/month for 40,000 executions with SSO and self-hosting support; Enterprise is custom. All cloud tiers include unlimited users, workflows, and integrations, with an annual-billing discount of about 17% versus monthly. The self-hosted Community Edition is free software with unlimited executions and all 500+ integrations — you pay only for your own server (from a few dollars/month on a basic VPS).

Typical workflows

• A DevOps team self-hosts n8n internally and builds a workflow that monitors error logs, uses an AI node to summarize and triage new incidents, and opens a ticket only when the summary matches specific severity criteria.
• A technical marketing team builds a multi-agent workflow where one AI agent researches a lead's company, a second drafts a personalized outreach email, and a human reviews the draft before it sends.

Integrations

• 500+ pre-built app integrations
• Generic HTTP request and webhook nodes for any API
• OpenAI, Anthropic, and other LLM providers, with native RAG and multi-agent support
• Model Context Protocol (MCP) support
• Git-based version control for workflows (Business/Enterprise tiers)

Privacy & security notes

Because n8n can be fully self-hosted, it's one of the strongest privacy options in this category for teams that need workflow data to stay entirely within infrastructure they control. n8n Cloud, by contrast, processes data on n8n's own servers like any hosted SaaS tool: n8n's published sub-processor list names Microsoft Azure (EU) and Hetzner (Germany) for hosting, and OpenAI, Anthropic, and Google Vertex AI for AI-related processing. n8n's privacy policy commits to retaining personal data no longer than necessary but does not publish a specific numeric retention period for Cloud workspace data — check n8n's Data Processing Agreement directly for that detail before running sensitive data through the hosted option.