What is n8n used for?

n8n is a visual workflow builder. You drag nodes onto a canvas and connect them: a trigger node starts the workflow (a webhook fires, a form gets submitted, a schedule ticks over), and downstream nodes act on that event — call an API, transform data, branch on a condition, write to a spreadsheet or database, send a Slack message or email. It ships with hundreds of pre-built app integrations plus generic HTTP request and webhook nodes, so if an app doesn't have a dedicated node, you can usually still connect to its API directly.

The self-hosted part matters for who this tool tends to attract. n8n's core is open-source, so it can run on your own server via Docker, giving you full control over where your data lives and what it costs at scale — relevant for anyone handling customer data under regional compliance requirements. There's also a hosted n8n Cloud option for teams that don't want to manage the infrastructure themselves, priced by usage.

The honest distinction worth making, especially against the backdrop of "AI agents": n8n executes the same steps in the same order every time, given the same trigger — that's what makes it reliable, auditable, and cheap to run at volume. An AI agent, by contrast, reasons about what to do next and can genuinely vary its path based on the situation, which is powerful for judgment calls but harder to fully trust running unsupervised. Calling n8n "AI" on its own oversells it; it's workflow plumbing, and that's a feature, not a limitation.

Where the two combine in real production systems: n8n handles the deterministic backbone — moving data between systems reliably, on schedule, with retries and error handling — and calls out to an LLM or a purpose-built agent only for the specific step that genuinely needs judgment, like drafting a reply or classifying an ambiguous request. That combination tends to be more trustworthy than routing an entire process through an agent, and cheaper to run than doing everything with LLM calls.

Typical real-world uses: syncing leads between an ad platform and a CRM, routing form submissions to the right team with enrichment along the way, generating scheduled reports by pulling data from several tools into one summary, and alerting a team in Slack or email when a metric crosses a threshold. The trade-off worth knowing upfront: complex branching logic has a real learning curve on the canvas, and self-hosting means you own uptime, backups, and updates — that's the price of owning your own automation instead of renting someone else's black box.