What is n8n?
n8n is a workflow automation platform that connects apps, databases, APIs and AI models through visual flows. You can run it on n8n Cloud or host it yourself, and it calls itself an AI workflow automation tool for technical teams.
Each workflow starts with a trigger, such as a webhook, schedule or new form entry, then passes data through nodes that read, transform or send it. The official site lists more than 500 integrations and over 10,000 templates. Where no node exists, you can call any API with an HTTP request or drop in JavaScript or Python. Recent releases put AI at the centre, with agent nodes, retrieval-augmented generation, MCP connections and evaluation tools.
The code is published on GitHub under the Sustainable Use License, which the repository describes as fair-code rather than open source. The homepage shows about 207,500 GitHub stars, a 4.7 out of 5 rating on G2 and more than 200,000 community members. The company also hosted an In The Loop conference in Berlin in October 2026, so the product is active and growing.
Key Features
AI agents with human approval steps and code-level control are the features that matter most in n8n, because they let you automate real processes without losing oversight.
- Visual workflow canvas: You chain triggers, actions and logic nodes on a drag-and-drop canvas and see each step’s input and output side by side. Single steps can be re-run on their own, which shortens debugging on long flows.
- AI agents and RAG: n8n lets you build single or multi-agent setups and retrieval-augmented generation systems inside a normal workflow. The canvas shows each reasoning step, and you can switch between cloud and local models without rebuilding the flow.
- Code where you need it: Any workflow can include JavaScript or Python steps alongside the no-code nodes. Developers use this to reshape data, call unusual APIs or add custom logic that a prebuilt node does not cover.
- AI Workflow Builder: You describe a workflow in plain language and the builder picks, places and configures the nodes. Each create or modify request uses one Assistant credit, and the docs say credentials are not sent to the language model.
- Guardrails and evaluations: Human-in-the-loop approvals and rule-based guardrails stop an agent before it takes a risky action. Built-in evaluation tools test AI steps against real data, so you can compare prompts or models before going live.
- Self-hosting options: The source code sits on GitHub, and you can deploy with Docker, on-premises or fully air-gapped. Queue mode for scaling is included in the free Community edition, while multi-main mode needs a paid licence.
Pros
Execution-based pricing is the main draw for n8n users, since a 20-step workflow run counts as one execution, not twenty billable tasks.
- Free to self-host: The Community edition costs nothing and n8n says you can stay on it indefinitely. Small teams with server skills can run unlimited workflows on their own hardware and pay only for hosting.
- Fair execution billing: n8n counts one execution per full workflow run, regardless of how many steps it contains. Long, multi-step automations therefore cost far less than on tools that charge for every task or operation.
- Control over data: Self-hosted and air-gapped deployments keep sensitive records inside your own network. This matters for regulated sectors and for AI workflows that handle customer or internal documents.
- Deep AI tooling: Agents, RAG pipelines, MCP connections, guardrails and evaluations all live in one product. Teams can build and test AI automations without stitching together separate agent frameworks and schedulers.
- Large community and templates: The homepage reports more than 200,000 community members and over 10,000 workflow templates. Beginners can copy a working template and adapt it rather than starting from a blank canvas.
Cons
Non-technical users find n8n harder to learn than simpler trigger-action tools, because data mapping and expressions assume some developer thinking.
- Learning curve: Concepts such as items, expressions and data mapping take time for people without a technical background. Starting from templates and the AI Workflow Builder helps, but complex flows still need some coding comfort.
- Licence limits on reuse: The Sustainable Use License allows internal business, personal and non-commercial use only. Agencies or software firms that want to resell or embed n8n should read the licence or ask about OEM terms first.
- Self-hosting takes work: Running your own instance means handling updates, backups, security patches and scaling. Teams without DevOps time may find the managed Cloud plans cheaper once staff hours are counted.
- Team features cost extra: SSO, Git version control, environments, log streaming and workflow sharing are missing from the free Community edition. The self-hosted Business plan that adds them is listed at €667 a month billed annually.
- Limited support below Enterprise: Starter, Pro and Business plans rely on the community forum for help. Dedicated support with a service-level agreement is reserved for Enterprise customers on custom contracts.
Who is Using n8n?
Technical teams in IT, DevOps, security and revenue operations form n8n’s core audience, alongside enterprises that need self-hosted AI automation.
- Operations and IT teams: They automate onboarding, access requests and internal ticket routing. Approval steps keep a person in control of sensitive changes.
- Developers and DevOps engineers: Engineers wire up alerts, deployments and lightweight internal tools. Code nodes and HTTP requests cover systems without a ready-made integration.
- Data and revenue teams: These teams sync CRMs, enrich leads and move records between databases. Workflows run on schedules or webhooks without manual exports.
- Large enterprises: The homepage shows logos including Vodafone, Mercedes-Benz, Deutsche Telekom and NVIDIA. Huel and Vodafone appear in case studies about saved hours and costs.
Pricing
n8n is free to self-host, while Cloud plans start at €20 a month billed annually (official pricing page, October 2026).
| Plan | Price | What you get |
|---|---|---|
| Community edition | Free | Self-hosted from GitHub, unlimited workflows, queue mode, no SSO, projects or Git version control |
| Starter (Cloud) | €20 per month, billed annually | 2,500 executions a month, unlimited users and workflows, 5 concurrent executions, 1,600 Assistant credits |
| Pro (Cloud) | €50 per month, billed annually | 10,000 executions a month, 3 shared projects, up to 50 concurrent executions, up to 9,600 Assistant credits |
| Business (self-hosted) | €667 per month, billed annually | 40,000 executions a month, 6 shared projects, SSO, environments and Git version control |
| Enterprise | Contact sales | Custom executions, 200+ concurrent executions, 365+ days of history, dedicated support with SLA |
The Cloud trial gives 1,000 executions and Assistant credits with no card required, and a 14-day Business trial is also offered. Companies with fewer than 20 employees may qualify for 50 per cent off Business. Futurepedia lists dollar prices of $24, $60 and $960 a month, which appear to be monthly-billed rates and do not match the euro annual figures above. One docs page also mentions a 14-day Pro trial, so trial terms may vary by region or date.
Disclaimer: execution quotas, credit allowances and overage fees change between plan updates, so confirm the figures on n8n’s pricing page for your currency and billing cycle.
What Makes n8n Unique?
Source-available self-hosting is the concrete difference: few major automation platforms let you run almost the full product on your own servers, even air-gapped, at no licence cost. Zapier and Make are cloud-only services, so data must pass through their infrastructure.
The second difference is depth for AI work. n8n treats agents as ordinary workflow steps, so a model can call tools, wait for a human approval, write to a database and log every prompt and response in one place. Combined with code in any step, this suits teams who outgrow no-code tools but do not want to maintain a custom agent stack.
Compatibilities and Integrations
More than 500 prebuilt integrations, plus HTTP requests and MCP, let n8n reach almost any system with an API.
- Deployment: n8n Cloud in the browser, or self-hosted with Docker, on-premises servers or air-gapped networks.
- Apps and data: Prebuilt nodes cover CRMs, databases, messaging apps, support desks and spreadsheets, with webhooks for anything else.
- AI models: Agent nodes work with major commercial model providers and local models, and connect to any MCP-compatible system.
- Developer access: A public API, JavaScript and Python code nodes, and Git-based version control on paid plans.
- Security tooling: SAML SSO, LDAP, external secret stores and log streaming to a SIEM on Business and Enterprise plans.
n8n Tutorials
A first n8n automation usually takes under an hour if you begin with a template rather than a blank canvas.
- Start a free Cloud trial on n8n.io, or install the Community edition with Docker on your own server.
- Open the template library and pick a workflow close to your goal, such as lead routing or an AI support agent.
- Add credentials for each connected app, then check every node’s input and output on the canvas.
- Use the AI Workflow Builder or a code node to adjust steps that the template does not cover.
- Add an approval step or guardrail before any action that sends emails, changes records or spends money.
- Activate the workflow and watch the execution log for errors during the first few days.
How We Rated It
Flexibility and self-hosting put n8n at 4.3 out of 5, while the learning curve and forum-only support on most plans keep it from a higher score.
| Criterion | Score | Why |
|---|---|---|
| Accuracy and Reliability | 4.3/5 | Retries and step-level logs |
| Ease of Use | 3.6/5 | Steep for non-technical users |
| Functionality and Features | 4.7/5 | Agents, code, RAG, evaluations |
| Performance and Speed | 4.2/5 | Queue mode scales self-hosting |
| Customization and Flexibility | 4.8/5 | Code in any node |
| Data Privacy and Security | 4.6/5 | Self-host, air-gapped option |
| Support and Resources | 3.9/5 | Forum only below Enterprise |
| Cost-Efficiency | 4.3/5 | Per-run, not per-step, billing |
| Integration Capabilities | 4.7/5 | 500+ integrations, MCP, HTTP |
| Overall Score | 4.3/5 | Average of the nine criteria |
Summary
n8n is one of the strongest choices for technical teams that want AI agents and business automation under their own control. Self-hosting, code in any step and per-run billing give it clear advantages over cloud-only rivals. The main limitation is the learning curve, and the licence restricts reselling or embedding without a separate agreement. Support below Enterprise also relies on the community forum. Start with the free Cloud trial and a template, then decide whether Cloud or self-hosting suits your team’s skills.

Reviews
Write a Review