n8n
A workflow automation platform for connecting apps and APIs, with visual building plus custom code when needed.
RecommendedLast reviewed July 17, 2026
What n8n actually is.
n8n is a workflow automation tool built for people who want the visual builder most of the time and raw code when the connector doesn't do the job. You drag nodes together to connect apps and APIs, then drop into JavaScript or Python inside any step once the built-in logic runs out. It ships as a cloud version and a self-hosted version you run on your own server via Docker.
The visual builder is easy to open and start dragging nodes into place. Getting a workflow production-ready is a different story: a node three steps into the chain that chokes on an unexpected API response can silently kill the whole run, and nothing retries it unless someone built that logic in first. n8n connects to more than 500 apps out of the box, but anything more delicate than a simple sync still needs someone who understands the logic underneath, not just the interface.
What it does
- Visual drag-and-drop builder for connecting apps and APIs.
- Code nodes let you write JavaScript or Python inside any workflow step.
- Self-hosts via Docker with full source code access on GitHub.
- Built-in support for multi-agent AI setups and RAG pipelines.
- MCP support connects workflows to outside AI tools and models.
- Webhooks and a REST API let finished workflows run headlessly.
- Over 500 pre-built app integrations, plus custom API connections for the rest.
Where it sits in the pipeline.
- AI & Intelligence
- Data & Enrichment
- Scraping & Custom Data
- Email Infrastructure & Deliverability
- Outreach & Sending
- LinkedIn & Multichannel
- Signals & Intent
- Calls & Meetings
- CRM & Pipeline Ops
Runs well with
The tools we'd run next to n8n, and what each pairing covers.
Pricing, without the fiction.
From the n8n pricing page · verified July 17, 2026
- 2,500 workflow executions/month
- 5 concurrent executions
- 1 shared project
- Unlimited users
- 50 AI Workflow Builder credits
- 10,000 workflow executions/month
- 20 concurrent executions
- 3 shared projects
- 150 AI Workflow Builder credits
- Admin roles, global variables, workflow history, execution search
- 40,000 workflow executions/month
- 6 shared projects
- SSO, SAML, LDAP authentication
- Git version control, scaling options, different environments
- Self-hosted deployment option
- Custom execution volume
- Unlimited shared projects
- 200+ concurrent executions
- 1,000 AI Workflow Builder credits (cloud only)
- Dedicated support with SLA, invoice billing
Self-hosting (Community Edition) is free and open-source, which is the real draw for agencies running workflows across many clients: cost doesn't scale with client count. The cloud version starts at Starter, €20/month billed annually, but caps out at 2,500 executions and 5 concurrent runs, with only 50 AI Workflow Builder credits included. Pro (€50/month) quadruples the execution cap and adds admin roles and workflow history. The real jump is Business at roughly €667/month, which is where SSO/SAML, Git version control, and self-hosted deployment options appear. Teams that self-host from the start avoid most of this ladder entirely and only pay for their own server.
Choose it, or skip it.
Choose it if
- Agencies running many client workflows: self-hosting is free (Community Edition), so headcount of clients doesn't move the bill.
- Technical builders who want JavaScript or Python nodes on hand instead of being boxed into no-code connectors for edge cases.
- Teams building AI agent workflows: multi-agent setups, RAG, human-in-the-loop steps, and MCP support are native to the platform.
- GTM teams with 500+ integration needs who'd rather build once and connect everything than pay per-task across separate point tools.
Skip it if
- Non-technical teams expecting a finished workflow same-day; the visual builder still takes days of iteration to get error handling right.
- Teams that won't self-host and need real execution volume: the jump from Pro to Business (self-hosted, SSO) lands around €667/mo.
- Buyers who want one dead-simple two-app zap with zero interest in touching a code node, ever.
If not n8n, then what.
- ZapierFree forever tier (100 tasks/mo, 2-step Zaps); paid from $19.99/mo (Professional)
Pick Zapier for a non-technical team that wants the largest app library and zero code, and will pay per task for it.
- Makefree tier (1,000 ops/mo) + paid plans from $12/mo (Core)
Pick Make for a similar visual-plus-logic approach with a lower cloud entry point if self-hosting isn't wanted.
- ApifyFree tier ($5/mo usage credit) then paid plans from $29/mo (Starter), usage-based overage billing
Pick Apify instead when the job is purely scraping at scale, not general cross-app orchestration.
Asked in real client work.
Is n8n actually free to use?
Yes, if self-hosted. The Community Edition is free with full source code access on GitHub and Docker deployment. The hosted cloud version starts at the Starter plan (€20/month billed annually) and adds execution caps, concurrency limits, and a set number of AI Workflow Builder credits per month.
What actually forces an upgrade on the cloud plans?
Two things: workflow executions per month (2,500 on Starter, 10,000 on Pro, 40,000 on Business) and concurrent executions (5, 20, then scaling further up). Shared projects, insights retention, and AI Workflow Builder credits also expand at each tier, so teams running frequent or parallel workflows hit the execution ceiling first.
How does n8n compare to Zapier or Make?
All three connect apps visually, but n8n lets you drop into JavaScript or Python inside a workflow instead of staying locked to pre-built connector logic, and it can be fully self-hosted for free. Zapier and Make are cloud-only and typically charge per task or operation, which scales cost with volume in a way self-hosted n8n does not.
Can n8n run AI agents, not just simple automations?
Yes. It supports multi-agent setups, RAG system construction, compatibility with multiple cloud or offline AI models, human-in-the-loop approval steps, and MCP (Model Context Protocol) support. Workflows can also be triggered and managed headlessly via webhooks and the REST API once built.
Is n8n hard to learn?
The visual builder itself is approachable and shows inputs/outputs at each step without much navigation. The real cost is in getting a multi-step workflow production-ready: error handling, retries, and edge cases typically take days of iteration rather than an afternoon, even for technical builders.