n8n

n8n is a source-available workflow automation platform where integrations are built in a visual editor backed by a node-based execution engine. It ships with more than 1,000 integrations for popular services and supports custom nodes for anything not covered, and it runs either as a hosted cloud offering or self-hosted on your own server.

Version 2.0, stable since December 2025, introduced draft and published workflow states, isolated task runners that are secure by default, database performance gains of up to 10x, OpenTelemetry tracing, and connections to MCP servers for working with AI agents. Workflows are stored as JSON, so they can be versioned like code.

A workflow is a flowchart that actually runs. Each node receives items from the nodes before it, does its job, and passes results onward, so data moves through the graph as a series of transformations. Trigger nodes start the run, action nodes call services or reshape data, and logic nodes like if and switch decide which path execution takes. Because every node operates on a stream of items, one run can process a single record or a thousand without changing the design.

Every automation needs a starting gun, and in n8n that is the trigger node. Schedule triggers fire on intervals or cron expressions for recurring work like nightly reports, webhook triggers expose an HTTP endpoint that outside systems call to start a run instantly, and service-specific triggers react to events such as a new database row or an incoming message. One platform covers both timed batch jobs and real-time, event-driven automation, and in version 2.0 a workflow goes live only when its published version says so.

// Code node: shape items before the next step
return items.map((item) => {
  return {
    json: {
      email: item.json.email.toLowerCase(),
      createdAt: new Date().toISOString(),
    },
  }
})

Sometimes no prebuilt connector does exactly what your process needs, and that is where the code node comes in. It runs custom JavaScript against the items flowing through a workflow: reshaping data, performing calculations, or calling endpoints that lack a dedicated node. In version 2.0, this code executes in isolated task runners, which keeps custom logic securely separated from the platform itself. The escape hatch means a workflow is never blocked by a missing integration.

We use n8n to automate the operational work that eats your team's week, connecting AI models, databases, and external APIs into workflows that run unattended. Workflows are exported as JSON and kept in version control, so every change is reviewable rather than an untracked edit in a live editor, and the draft and published states in 2.0 mean updates never break a running automation. AI steps call the OpenAI API to classify, summarize, or draft content inside a larger process. Instances are self-hosted in Docker, so your data never leaves infrastructure you control.

Self-hosting means the automation platform runs on your servers, under your rules. The official Docker image packages n8n for consistent deployment, a persistent database holds workflow definitions, credentials, and execution history, and environment variables configure authentication and webhook URLs. Version 2.0 sharpens this story with task runners isolated by default and database performance up to 10x faster, while OpenTelemetry tracing gives operations teams real visibility into every run. We deploy n8n behind a reverse proxy with TLS, in containers alongside the rest of your system.