n8n is not behaving like a static automation product. ToolVitals logged 46 release events in 30 days, 30 GitHub releases in 90 days, and the latest public releases are full of AI-builder and editor fixes. The signal is clear: this team is spending real effort on AI workflow control, not just shipping another visual glue tool.
The public site backs that up. n8n now describes itself as “AI agents and workflows you can see and control,” pushes visual building plus code, and says you can deploy on your own infrastructure or theirs. It also claims 500+ integrations on the site, while the ToolVitals payload still lists 400+, so treat the integration count as a moving marketing target, not a fixed fact.
The release notes show where the energy is going. n8n@2.18.5 focused on AI-builder cleanup, including hiding intermediate AI-created workflows, fixing the AI view stacking context, and tightening version-menu behavior. n8n@2.19.0, marked pre-release, added credential auth and tests for Twilio, Pipedrive, Asana, Freshdesk, and Workable, while also shipping more ai-builder fixes. That is not random churn. It looks like a platform hardening its AI surface area so people can trust it in production.
The blog post from April 29 makes the strategy even more obvious. n8n is actively publishing guidance on human-in-the-loop versus human-on-the-loop systems, with an explicit focus on latency, auditability, and compliance. That lines up with the product page, which emphasizes humans in the loop, structured outputs, audit trails, RBAC, workflow diffs, and on-prem deployment. The bet here is not “let the model run wild.” The bet is “let the model run, but keep it observable and governable.”
What this data does not tell you
ToolVitals can see release cadence, stars, and recent activity. It cannot see code quality, customer satisfaction, revenue, support load, or whether the product actually behaves well under production traffic. A busy release log can mean serious momentum, or it can mean a product that needs constant repair. The metrics alone do not tell you which one this is.
It also does not tell you whether the AI features are actually useful, or whether teams are adopting them beyond demos. The public messaging is strong. The proof of value still lives outside the dataset.
Compared with nearby tools
The closest automation peers in the related data are moving too, but on a smaller scale. Skyvern has 21 release events in 30 days and 21,401 GitHub stars. Sim has 30 release events in 30 days and 27,880 stars. n8n sits at 46 release events in 30 days and 185,884 stars. That is a different league of attention and a much heavier shipping tempo.
Bottom line
If your team needs self-hosted workflow automation with AI steps, human approval gates, and strong governance language, n8n deserves a pilot now. If you just need a simple automation runner, this is probably too much product. The current evidence says n8n is betting hard on controlled AI workflows, and it is shipping like it means it.
Sources: https://n8n.io, https://github.com/n8n-io/n8n, https://github.com/n8n-io/n8n/releases/tag/n8n%402.18.5, https://github.com/n8n-io/n8n/releases/tag/n8n%402.19.0, https://blog.n8n.io/human-in-the-loop-vs-human-on-the-loop/