GILD verdict
Choose n8n when technical teams want automations and AI agents under their own control.
Settle hosting, the license fit, and who owns each workflow before business processes depend on it.
n8n
Workflow Automation by n8n
0 MixedFit and limits
Who it fits
- Your automations touch data that must stay on your own infrastructure: the Community edition runs self-hosted with Docker or Docker Compose, on premises or in a private cloud.
- Your workflows are long or branch heavily: n8n bills one execution per production run, whatever the number of steps, and sub-workflows and error workflows do not count.
- Someone on the team writes code: Code nodes run JavaScript and Python, and the HTTP Request node reaches any API without a prebuilt integration.
- You want AI agents inside workflows with controls around them. The AI Agent node calls tools you connect, human approval can gate individual tools, and the Guardrails node checks text before and after the model.
Where it falls short
- The Sustainable Use License limits the free edition to internal business use: you may not host n8n as a service or let outside users build or edit workflows without a commercial agreement.
- Git source control, environments, and SSO need a Business or Enterprise license, and external secret stores need Enterprise. n8n's security page places SSO on Enterprise alone, so confirm SSO in the Business quote. In a case study n8n published, Stepstone reports that hundreds of workflows on one shared Community account became unsustainable, and the company moved to Enterprise for version control and separate environments.
- Self-hosting makes your team the operator: database, updates (n8n recommends at least monthly), task runner isolation for Code nodes, and queue mode with Redis to scale.
- On n8n Cloud, Starter caps an execution at 5 minutes and 5 concurrent runs, and Business, the first plan with Git and SSO, is not offered as a hosted plan.
Implementation reality
Choose the edition and the host
Decide first whether n8n Cloud or self-hosting fits your data rules. Cloud plans run in Frankfurt and n8n manages updates on a Stable or Beta track. Self-hosted, the free Community edition covers internal use; registering it by email adds folders and debugging in the editor, and a Business or Enterprise key adds SSO, Git, and environments. One n8n Community member running 8 workflows reports a US$6-a-month Hetzner server and 1 to 2 hours a month of updates, backups, and log checks.
Count executions before you buy
Estimate monthly production runs. A schedule every 5 minutes produces about 8,600 to 8,900 runs a month. Each webhook request counts once. Starter includes 2,500 executions and Pro 10,000. Self-hosted Business starts at 40,000, with overage of €4,000 per extra 300,000.
Give every workflow an error path
Set an error workflow in every production workflow's settings. Start it with the Error Trigger and send failures to Slack or email. Use Stop And Error when bad data must fail a run and trigger the same alert.
Bound what agents can do
Connect only required tools. Require human approval for messages and record changes, including deletions. Run known test cases before changing the model or prompt. n8n labels its Agents builder Preview and warns that its behavior may change. Keep business-critical agents on the AI Agent node.
What to measure in your pilot
- GILD found upkeep figures only from solo self-hosters in n8n's community forum; log server spend and admin hours each month, and price n8n Cloud once upkeep costs more than its plan.
- Track how many people edit shared production workflows each month, and add Git and a staging environment before a second team starts editing them.
- Rerun the same agent test cases on the Agents builder after each n8n release, and move a business-critical agent there only once its results match the AI Agent node.
Sources
- Official documentation
- 22
- Public source
- 1
- n8n pricing
- n8n docs: Understand executions
- n8n docs: Community license
- n8n docs: License FAQ
- n8n docs: Host n8n
- n8n docs: Compare editions
- n8n docs: Understand concurrency (Cloud)
- n8n docs: Update your version (Cloud)
- n8n docs: Handle errors gracefully
- n8n docs: Use source control and environments
- n8n docs: Configure SSO
- Security at n8n
- n8n docs: Use external secret stores
- n8n docs: Update n8n
- n8n docs: Set up task runners
- n8n docs: Enable queue mode
- n8n docs: AI Agent node
- n8n docs: Build and manage agents
- n8n docs: Human-in-the-loop for tools
- n8n docs: Guardrails node
- n8n docs: Test and improve AI workflows
- n8n case study: How Stepstone moved from four early adopters to a 3,000-employee rollout
- n8n Community: n8n vs Zapier, my honest cost breakdown after 1 year of self-hosting (Apr 11, 2026)