n8n

n8n (pronounced "n-eight-n," short for "nodemation") is an open-source workflow automation platform offering the same visual, node-based automation-building experience as Zapier or Make, but with a key structural difference: it can be self-hosted on a company's own infrastructure (or used as n8n's managed cloud offering), giving builders full control over data residency and eliminating the per-task/per-operation billing model that makes high-volume automation on Zapier or Make expensive at scale. Why it matters: n8n has become the go-to choice for technically capable teams — often those with at least one developer comfortable with basic server maintenance — who need Zapier-level automation flexibility but either can't send certain data to a third-party SaaS platform (data residency, compliance, or security requirements) or run high enough automation volume that per-task pricing becomes a significant recurring cost. It's also popular in the AI-agent-building community specifically, because n8n has invested heavily in native AI/LLM nodes (calling models like Claude or GPT directly, chaining multiple AI calls, building basic agentic workflows) alongside its traditional app-to-app automation capability, making it a common choice for builders prototyping AI-powered automations without writing a full custom backend. How it works: like Make, n8n represents workflows as a node-based canvas — nodes for triggers (webhooks, schedules, or app-specific events), actions (HTTP requests, database operations, or 400+ pre-built app integrations), and logic (IF/Switch nodes for branching, Merge nodes for combining data streams, Loop nodes for iterating over arrays). Self-hosting typically means running n8n via Docker on a VPS or Kubernetes cluster, connecting it to a Postgres database for workflow storage, and managing your own uptime/backups — trading the "someone else's problem" simplicity of a SaaS platform for full control and (usually) significantly lower cost per execution at scale. Worked example — a self-hosted n8n workflow that classifies and routes support emails using AI: Node 1 (Trigger): "IMAP Email" node watches a support inbox for new messages. Node 2 (AI node): sends the email body to an LLM with a prompt "Classify this support email into one of: Billing, Technical, Sales, Other. Respond with only the category." Node 3 (Switch node): branches based on the AI's classification — Billing routes to a finance Slack channel, Technical creates a ticket in a self-hosted Zammad instance, Sales notifies an AE, Other gets a generic auto-reply. Because this runs entirely on the company's own server, the potentially sensitive email content never passes through a third-party automation vendor's servers — a common deciding factor for teams choosing n8n over Zapier or Make for anything touching regulated or sensitive customer data.