n8n

n8n sits in the middle of the automation market: more powerful than Zapier, less polished than Make, and genuinely free when you run it yourself. The open-source Community Edition gives you unlimited workflow executions with no usage caps, which is the key reason technical teams keep choosing it over cloud-only competitors once their automation volume grows.

The visual canvas lets you build workflows by connecting nodes, each representing a trigger, action, or transformation. You can drop into JavaScript or Python at any point, which means there’s no ceiling on what you can do. Over 500 native integrations cover the usual suspects (Slack, Postgres, HTTP requests, webhooks), and the built-in AI nodes let you wire up OpenAI or Anthropic calls directly into workflows without a separate integration layer.

Where n8n has pulled ahead recently is in AI agent building: multi-step reasoning chains, tool-use patterns, and human-in-the-loop approval steps all have first-class support on the canvas. The real-time execution inspector lets you see exactly what each node received and returned, which makes debugging far less painful than working blind.

The honest limitation is that self-hosting n8n is a real commitment. You’ll want Docker familiarity, and you should budget two to four hours a month for updates and occasional troubleshooting. If webhooks and JSON structures feel foreign to you, there’s a learning curve before you become productive. The cloud plans exist for teams who want to skip the infrastructure work, but the execution limits on entry-level tiers run out fast if you’re using polling triggers. Railway is a middle path: your own n8n instance on managed infrastructure, no execution caps, and your data off n8n’s servers.

For a developer, indie hacker, or small technical team with even modest automation volume, self-hosting n8n is almost certainly cheaper and more capable than any SaaS alternative.

n8n: Pros & Cons