n8n-io/n8n

Worth a serious look if you need to automate multi-app business processes and want the runs to stay on infrastructure you control. Treat the license, not the feature list, as the main decision point.

The problem

Say a team needs form submissions to land in a CRM, get enriched, trigger a Slack message, and wait for human approval before an invoice is issued. Doing this by hand means writing glue scripts, scheduling them, storing credentials, and building some dashboard to see whether a run failed halfway. With a visual automation platform the same flow becomes a canvas of connected nodes: a trigger, integration steps, a branch, and an approval gate. The change is not just speed of the first build; the workflow itself becomes the documentation of what happens and where it stops.

What's different

The distinctive bet here is combining a node canvas with real code. The README says JavaScript, Python and npm packages can be dropped into a workflow, so the visual layer handles routing and integrations while an awkward transformation goes into a code node rather than a separate service. On the AI side, the project presents model choice as a configurable step — OpenAI, Anthropic, Google or open-source models — so swapping providers does not mean redesigning the flow AI guide. It also supplies a large integration catalogue and template gallery, which shortens the blank-page problem for common patterns.

The trade-off is that this is a platform, not a library. You inherit its execution model, its upgrade cadence and its licensing. The licence is the sharpest edge: n8n describes itself as fair-code under the Sustainable Use License plus an enterprise licence, with source available and self-hosting allowed but not OSI open source project instructions licence terms. The 2022 HN thread's title accused the project of not being open source; the founder replied that the debate led to fair-code and later the Sustainable Use License founder reply. A later 2025 commenter noted the docs now use that framing and asked whether the complaint was still relevant HN comment. For a company, that is a procurement question to settle early, not a footnote.

Who should try it

Good fit: an operations or platform team with many SaaS tools and a few people who can read code, who want a shared, self-hosted place to build and inspect automations. It also suits someone prototyping an AI-assisted back-office flow with human approval steps, because the visual structure makes the handoffs visible.

Poor fit: a developer who just needs one scheduled script and version control, or anyone who requires an OSI-approved licence. Also a poor fit for a team that will not staff ongoing platform upkeep; a running automation server is infrastructure, with credentials, upgrades and failed executions to watch.

The cloud login is a documented alternative to self-hosting, which changes who can start without Docker knowledge cloud option.

Requirements and limitations

The quick-start path assumes Docker, so self-hosting means operating a container plus a persistent data volume Docker instructions. That volume matters: workflows, credentials and execution history live there, and you need a backup and upgrade routine for it. Integrations often need OAuth apps or API keys of your own, so expect setup work per service, not zero-config connection. The enterprise features and support sit behind separate commercial licences, so check which capabilities you actually need before assuming the free path covers them enterprise licence.

HotRepo's take

HotRepo's dated snapshots show 201,645 stars on 2026-08-22 and 208,850 on 2026-10-28, with five daily-board appearances in that window project page. The sample is too sparse to judge momentum; it only shows the project remained large and active across those observations. Before adopting, verify the Sustainable Use Licence against your intended commercial use, because that clause decides more than any feature comparison.