n8n Review 2026

n8n is a workflow automation platform where you wire nodes together on a visual canvas to connect apps, transform data, and trigger actions.

It differs from Zapier in three structural ways. First, the license is fair-code under n8n Sustainable Use License, which means you can self-host the full product for free on your own server but cannot resell it as a hosted service. Second, nodes can contain JavaScript or Python code, so power users drop custom logic inline instead of maintaining external scripts. Third, the platform ships with 70+ LangChain-based AI nodes for agents, vector stores, memory, and LLM calls.

That combination, self-host freedom, inline code, native AI, is why technical founders and AI builders picked n8n as their default automation tool in 2025 and 2026.

The core workspace is the workflow editor, a node-based canvas where you drag triggers, actions, and transforms onto a grid. Every workflow starts with a trigger, a webhook, a schedule, a database polling node, or a chat trigger for AI agents. You wire the trigger to action nodes for Slack, Google Sheets, Airtable, HubSpot, or any of the 500+ integrations. Between nodes you drop transforms. Set, Code, If, Switch, or Merge to shape data.

The AI subsystem is the 2026 standout. The AI Agent node runs LangChain tool agents that can call any other n8n node as a tool, maintain conversation memory, and stream responses. Vector Store nodes connect to Pinecone, Qdrant, Supabase, and Azure AI Search for retrieval-augmented generation. The Chat Trigger node turns any workflow into a chatbot endpoint.

Execution history is a debugging lifeline. On Pro and above you get workflow history, which lets you replay a failed execution with the original input data. Below Pro you get 7 days of execution logs with basic replay, which is enough for hobby projects but tight for production.

n8n offers six tiers plus the free self-hosted option.

Community Edition, $0, self-hosted with all 500+ integrations, unlimited executions, unlimited workflows, no SSO or RBAC, support via forum only.

Cloud Starter, 24 EUR per month (20 EUR annual), 2,500 executions, 5 active workflows, 7 days of history. Fine for one-off projects or early-stage validation.

Cloud Pro, 60 EUR per month (50 EUR annual), 10,000 executions, 15 active workflows, 14 days of history. This is the plan most small teams standardize on.

Cloud Pro 50K, 120 EUR per month (100 EUR annual), 50,000 executions, 50 active workflows, for scaling teams.

Business, 800 EUR per month (667 EUR annual), 40,000+ executions, unlimited active workflows, SSO, SAML, LDAP, advanced RBAC, audit logs, Git versioning, 365 days of history. Startups under 20 employees qualify for a 400 EUR per month rate.

Enterprise, custom pricing, dedicated CS, 99.9% SLA, self-hosted enterprise license key with SSO and Git.

Cloud plans bill in EUR. Self-hosting is free software, you still pay server costs, budget $5 to $50 per month for a small VPS.

The sticker price tells only part of the story with n8n because the execution model compounds differently than Zapier.

On Cloud Pro at 60 EUR per month, a moderately active workflow running every 15 minutes (96 executions per day, 2,880 per month) leaves headroom for about 2 more similarly active workflows before hitting the 10,000 cap. Teams that mix long-running AI agents with scheduled data syncs often graduate to Pro 50K (120 EUR per month) within 3 to 6 months.

Self-hosting trades subscription spend for infrastructure and operational overhead. A single 2 GB VPS runs around $10 to $20 per month and handles most small team workloads. Once you cross roughly 10 active workflows with frequent triggers, plan for a 4 to 8 GB instance at $30 to $50 per month, plus database and backup storage. Realistic total cost for a self-hosted production setup is $400 to $800 per year.

The hidden cost is engineering time. Expect 10 to 20 hours for initial setup (Docker, reverse proxy, SSL, backups) and 2 to 5 hours per month for updates, monitoring, and debugging. At a $100 per hour blended engineering rate, self-hosting saves money versus Cloud Pro only above roughly 40,000 executions per month.

Capterra averages 4.9 out of 5 across published reviews, G2 sits at 4.5 with users calling out customization depth and AI node quality. Trustpilot sentiment is more mixed, with the same split we see for Ahrefs, power users on higher tiers love the product, casual users on Starter struggle with the learning curve.

The most common complaint on Reddit and the n8n community forum is debugging. Errors often surface as empty output from a downstream node with no clear cause, and the debug panel requires data structure fluency to interpret. Users who came from Zapier describe the first month as the hardest, followed by a second month where productivity jumps past what Zapier allowed.

The most common praise is the combination of self-hosting and AI nodes. A typical G2 review reads like, we replaced Zapier, our internal scripts, and part of our LangChain backend with one n8n instance. That consolidation value is what drives the 30% commission recurring affiliate economics, customers who commit rarely churn.

Support tiers mirror the pricing. Community and Starter users get forum-only help, with n8n staff and community experts answering questions on the community.n8n.io forum and GitHub issues. Pro users unlock priority forum support and email ticketing. Business and Enterprise customers get direct email and in some cases dedicated CS.

Response times on the forum are fast, typically under 24 hours for technical questions thanks to an active community. Paid email support replies within business hours Monday through Friday, Central European Time, which is a timezone gap if your team runs US business hours. There is no phone support on any plan.

Documentation at docs.n8n.io is comprehensive on nodes and expressions but lighter on end-to-end recipes. The gap is filled by YouTube creators and the n8n workflow template library at n8n.io/workflows, which hosts several thousand community-submitted automations.