What Is n8n? Complete Guide to Source-Available Workflow Automation (2026)

n8n (short for “node-based no-code” a numeronym of the word “nodemation”) is a fair-code, source-available workflow automation platform that lets you connect apps, APIs, databases, AI models, and internal systems using a visual node-based editor. Its free self-hosted Community Edition sits on GitHub with over 190,000 stars, and its cloud-hosted paid plans run on execution-based pricing starting at �20/month.

“Tools like ChatGPT and Claude are great, but n8n is the thing that allows you to integrate AI into your work and processes in a safe and controlled way.”

Ollie Scheers, CTO at Huel

n8n is not traditional OSI open source. It follows the Sustainable Use License, a fair-code model that makes the full source code public and modifiable but restricts reselling, embedding n8n into commercial products, or hosting it as a paid service for third parties. For internal company automation, the license distinction rarely matters. For commercial redistribution, it absolutely does.

What n8n Actually Does

n8n automates workflows sequences of connected nodes that each perform a discrete operation. A trigger node starts the chain (webhook, schedule, form submission, app event), and downstream action nodes read, transform, route, or write data.

Here is what n8n workflows do in practice across organizations in 2026:

Route leads from web forms into CRMs with enrichment and scoring
Sync records bidirectionally between tools that have no native integration
Send Slack or Teams alerts when database conditions change
Orchestrate AI agents that classify, summarize, draft, and route support tickets
Pull data from internal APIs behind private networks
Transform and load data into warehouses (ETL-style pipelines)
Trigger multi-step actions from webhooks or scheduled cron jobs
Run JavaScript or Python code inline for custom logic
Coordinate RAG (retrieval-augmented generation) workflows across LLMs and vector stores

n8n occupies the space between pure no-code tools like Zapier and custom backend engineering. Non-technical users build visually; engineers drop into code, self-host, extend nodes, and manage workflows through Git.

Core Features That Define n8n in 2026

Visual node-based editor: Drag-and-drop workflow canvas with full visibility into every step. Branching, loops, IF/ELSE conditions, parallel paths, sub-workflows, and error workflows are all first-class constructs.
500+ prebuilt integrations: Native nodes for Slack, Gmail, Google Sheets, Notion, Airtable, PostgreSQL, GitHub, Jira, HubSpot, Salesforce, and hundreds more.
AI-native toolchain: Built-in AI Agent node, text-to-workflow AI Workflow Builder, native LLM connectors (OpenAI, Anthropic, Google Gemini, Mistral, Cohere, Hugging Face, Perplexity, Jina AI), MCP (Model Context Protocol) support, guardrails for AI outputs, and evaluation nodes for testing AI accuracy.
Code nodes (JavaScript and Python): Drop into code anywhere in a workflow. Manipulate data, call undocumented APIs, implement custom business logic without leaving the editor.
HTTP Request node: Connect to any REST API or GraphQL endpoint even when no prebuilt node exists.
Self-hosting: Deploy via Docker, Kubernetes, or on bare metal. Run workflows entirely inside your own infrastructure behind firewalls, in air-gapped environments, on-premises, or in private clouds.
Git-based version control and environments: Push/pull workflows between dev, staging, and production. Compare diffs, review changes, and roll back the same way you manage application code.
Error workflows and execution debugging: Pin data, replay individual nodes, inspect inputs/outputs at each step, and attach error workflows that fire when executions fail.
Role-based access control (RBAC): Instance admins, project admins, project editors, project viewers. SAML/OIDC/LDAP for SSO. MFA enforcement. Audit logs with log streaming to tools like Datadog.
Execution-based pricing: You pay per successful workflow execution not per step, not per “operation.” A 50-step workflow and a 3-step workflow cost the same per run.

How to Build a Workflow in n8n

Every n8n workflow follows a consistent structure. Here is the standard build sequence:

Pick a trigger: Choose from webhook (incoming HTTP POST), schedule (cron), app event (new Gmail email, new Airtable row, new GitHub issue), manual trigger, form submission, or SSE stream. The trigger determines when the workflow runs.
Configure data input: The trigger delivers a payload a JSON object with fields you can reference in every downstream node. The webhook body, the email content, the form fields.
Add processing nodes: Route data through filters (IF/Switch), transformations (Set/Function), merges, loops (Split in Batches), date formatting, regular expression extraction, or code nodes. This layer handles business logic.
Connect action nodes: Write to a CRM, send a Slack message, update a database, call an external API, generate an AI summary, create a Notion page, send an email, or trigger another workflow. Action nodes are where work actually happens.
Define error handling: Attach an Error Trigger node and build a parallel error workflow. Send alerts to an owned Slack channel, log failures to a database, notify the workflow owner by email. Silent failures are organizational risk.
Test with real data: Use the “Execute Node” button to run individual nodes. Pin data to freeze intermediate outputs. Replay partial executions. Check inputs/outputs in the sidebar without leaving the canvas.
Activate and monitor: Switch the workflow from inactive to active. Monitor execution history, set up Insights dashboards, and optionally stream logs to your observability stack.

n8n vs Zapier vs Make: Comparison Table

Feature	n8n	Zapier	Make (Integromat)
License model	Fair-code (Sustainable Use License)	Proprietary, closed-source	Proprietary, closed-source
Self-hosting	Full Docker/Kubernetes support	Not available	Not available
Code support	Native JavaScript and Python nodes	Limited code steps (Python/JS)	Limited code modules
AI capabilities	AI Agent builder, LLM nodes, text-to-workflow, MCP, guardrails, evaluations	OpenAI and basic AI integrations via apps	Limited native AI; HTTP module workarounds
Visual editor	Full node-based canvas, branching, sub-workflows, parallel paths	Linear zap builder (trigger ? actions)	Scenario-style visual designer
Version control	Git-based push/pull, diff, environments (dev/staging/prod)	Not available	Not available
Custom nodes/plugins	Build via SDK, publish privately or publicly	App-based closed ecosystem	Custom modules via SDK
Pricing model	Per workflow execution (unlimited steps)	Per “task” (each action = 1 task)	Per “operation” (each module execution)
Error handling	Dedicated error workflows, per-node replay	Basic error notifications	Error-handling routes
SSO/SAML/LDAP	Business and Enterprise plans	Enterprise plans only	Enterprise plans only
Community size	200k+ members, 190k+ GitHub stars, 4.9/5 G2	Large non-technical user base	Active EU-centric community
Best for	Technical teams, AI builders, regulated industries, complex workflows	Non-technical users, simple automations, broad app coverage	Operations teams, visual thinkers, mid-complexity workflows

Cloud vs Self-Hosted: Choosing Your Deployment Model

n8n Cloud (Managed Hosting)

n8n Cloud is fastest for getting started. No infrastructure setup, no server maintenance, no database tuning. Pay monthly, build workflows immediately, access the full node library. Hosted in Frankfurt, Germany (EU), with US hosting available. SOC 2 compliant. Good for startups, solo builders, and teams without dedicated DevOps.

Cloud plans (2026):

Plan	Monthly Price	Executions	Key Additions
Starter	�20/mo	2,500	1 shared project, 5 concurrent, 50 AI Builder credits
Pro	�50/mo	10,000	3 shared projects, 20 concurrent, insights, workflow history
Business	�667/mo	40,000	6 shared projects, SSO/SAML/LDAP, Git versioning, environments
Enterprise	Custom quote	Custom	Unlimited projects, 200+ concurrent, log streaming, SLA support
Start-up Plan	50% off Business	40,000	For companies under 20 employees

Self-Hosted n8n (Community Edition)

The Community Edition is free, downloadable from GitHub, and gives you the full platform. Deploy via Docker with a single command, or orchestrate through Kubernetes for production scale. You own updates, backups, monitoring, SSL/TLS configuration, database health, and incident response.

Self-hosting is the right call when:

Your systems sit behind private networks or in air-gapped environments
Data residency regulations (GDPR, HIPAA) require on-premises processing
Credentials and API tokens must never leave your infrastructure
You process high volumes where per-execution cloud pricing becomes prohibitive
Your team already maintains production infrastructure and knows how to secure it
n8n’s documentation explicitly recommends self-hosting only for users with technical expertise misconfigured instances can cause data loss, security breaches, or downtime

The License: What “Fair-Code” Actually Means

n8n uses the Sustainable Use License and n8n Enterprise License. It is not OSI-approved open source, and n8n itself says it does not call itself open source.

The Sustainable Use License allows:

Using n8n for internal business operations (sync company data, automate internal processes)
Building integrations or custom nodes for your own product
Modifying the source code for personal or internal use
Redistributing code with the same license and attribution
Providing consulting services around n8n deployment and workflow design

The Sustainable Use License does not allow:

White-labeling n8n and selling it to customers as a hosted service
Hosting n8n and charging third parties to access it
Embedding n8n into a commercial SaaS product where n8n is the core value proposition

If your use case involves reselling, embedding, or operating n8n as a multi-tenant service for paying customers, contact n8n for an Enterprise license or consult your legal team.

AI and Agentic Automation in n8n (2026)

The biggest evolution in n8n through 2026�2026 has been its pivot toward AI-native workflows. The platform now includes:

AI Agent node: Build context-aware agents with memory, tools, and reasoning. Agents can query databases, call APIs, search knowledge bases, and escalate to humans all on the same canvas.
AI Workflow Builder: Describe a workflow in natural language (“Watch for new Gmail emails, summarize them with OpenAI, and post to Slack”) and n8n auto-generates the node chain. You refine instead of starting from scratch.
Native LLM connectors: OpenAI (GPT-4o, assistants, embeddings), Anthropic (Claude), Google Gemini, Mistral AI, Cohere, Hugging Face, Perplexity, Moonshot Kimi, Alibaba Cloud Model Studio, and more.
MCP (Model Context Protocol) support: Connect n8n to any MCP-compatible tool or data source, future-proofing AI workflows against model and provider changes.
RAG (Retrieval-Augmented Generation): Combine vector stores, embedding models, and LLMs to build domain-aware AI pipelines.
Guardrails and evaluations: Define structured input/output schemas for AI steps. Evaluate AI accuracy with built-in evaluation nodes before shipping to production.
Human-in-the-loop: Pause AI workflows for human approval at any step. AI classifies and drafts; humans approve and act.

Real-world example: Vodafone uses n8n for threat intelligence automation and reports �2.2 million saved. Huel uses n8n to build an “AI-first company culture” and saved over 1,000 hours of manual work. Meta, Microsoft, SEAT, Wayfair, Zendesk, and Delivery Hero are among the companies that have publicly adopted n8n.

Common Mistakes When Adopting n8n

Starting with irreversible workflows: Your first workflow should save time, not control billing, refunds, account suspension, or customer-wide email blasts. Build trust before automating actions that cannot be undone.
Forgetting workflow ownership: Every active workflow needs a named owner, a documented purpose, and a failure notification channel. Unclaimed workflows are the most common source of silent automation failures.
Assuming self-hosting is automatically better: Self-hosting is powerful but only if your team can secure, monitor, back up, and update the instance. n8n’s documentation explicitly warns that self-hosting mistakes can cause data loss and security breaches.
Hard-coding secrets and business rules: Use the encrypted credential store. Use environment variables or external secret stores (HashiCorp Vault) for production. Future you or the person inheriting the workflow will not remember the context.
Ignoring API deprecations: If an external tool changes or deprecates an API endpoint, connected workflows break silently. Build monitoring around critical workflows.
Designing overly large single workflows: Break complex logic into sub-workflows and use the Execute Sub-workflow node. Smaller workflows are easier to test, debug, and reuse.

n8n Readiness Checklist

Before adopting n8n as your team’s automation backbone, answer each question:

Who will build the workflows, and what technical skills do they have?
Who owns each workflow after launch? Who gets paged when it breaks?
Which workflows are low-risk enough to serve as your first projects?
Do you need Cloud (faster start, managed) or self-hosted (infrastructure control, data locality)?
Do you understand the Sustainable Use License for your intended use case?
What credentials and API tokens will n8n hold? How will those be rotated?
How will workflow failures be detected and who will respond?
How will workflow changes be reviewed before deployment PRs, diffs, staging environments?

If you can answer every question, n8n will likely become a reliable automation layer rather than another unmanaged tool. The acid test: build a real workflow, break it deliberately, and confirm your team can diagnose and fix it within the response window you committed to.

FAQ

Is n8n open source?

No not under the OSI definition. n8n is source-available and fair-code under the Sustainable Use License. The full source code is public and modifiable, but the license prohibits reselling n8n-as-a-service and embedding it into commercial products without an Enterprise license. n8n itself states it does not call itself open source.

Can I self-host n8n for free?

Yes. The Community Edition is free on GitHub and includes the full platform. Deploy via Docker. You handle your own infrastructure, security, backups, and updates. n8n recommends self-hosting only for users with server administration experience.

Do I need to know how to code?

Not for basic workflows. The visual editor and 500+ prebuilt nodes cover common automations without code. Coding helps for custom API calls, complex data transformations, conditional branching, error handling, and building private nodes but you can go far without it.

How does n8n pricing compare to Zapier or Make?

n8n charges per workflow execution (one full run of your workflow, regardless of how many steps it contains). Zapier charges per task (each individual action counts as a task). Make charges per operation (each module execution). A 20-step n8n workflow costs one execution; the same workflow on Zapier could cost 20 tasks. This makes n8n significantly cheaper for complex automations.

What are the actual 2026 prices?

Cloud plans start at �20/month (Starter: 2,500 executions), �50/month (Pro: 10,000 executions), �667/month (Business: 40,000 executions), and custom pricing for Enterprise. The self-hosted Community Edition is free. A Startup plan offers 50% off Business for companies under 20 employees.

Is n8n good for AI workflows?

Yes increasingly so. n8n includes native AI Agent and LLM nodes (OpenAI, Anthropic, Google Gemini, Mistral, Cohere, and others), a text-to-workflow AI builder, MCP support, guardrails, and evaluation nodes. It is being used for RAG pipelines, AI support triage, sentiment analysis, automated classification, and multi-agent orchestration.

Is n8n secure for production use?

Yes, especially when self-hosted. n8n Cloud is SOC 2 compliant and GDPR-ready, with data hosted in Frankfurt, Germany. Enterprise features include SSO/SAML/LDAP, RBAC, audit logs, log streaming to SIEM tools, encrypted credential storage, external secret store integration, and MFA enforcement.

What companies use n8n?

Public adopters include Meta, Microsoft, Vodafone, Zendesk, Wayfair, SEAT, Mistral AI, Paddle, Delivery Hero, Unbabel, Dedatech, and Onfleet. Vodafone reports �2.2 million in savings from threat intelligence automation. Huel saved 1,000+ hours of manual work through AI-first workflows.

Sources

n8n Official Website
n8n Features Overview
n8n Pricing Page
n8n Documentation: Sustainable Use License
n8n Documentation: Self-Hosting Guide
n8n Documentation: Error Handling
n8n GitHub Repository (190k+ stars)
n8n Blog: Re-learning AI Agent Development Tools in 2026
HatchWorks AI: n8n Guide 2026
Upskillist: n8n Review 2026
n8n Community Forum (200k+ members)
n8n Case Study: Vodafone
n8n Case Study: Huel
n8n G2 Reviews (4.9/5 rating)

What Is n8n? Complete Guide to Source-Available Workflow Automation (2026)