n8n Review (2026): The AI Assistant Built a Working Automation From One Sentence

We signed up for n8n Cloud’s 14-day trial (no credit card required) and gave its AI assistant one sentence: watch this site’s own RSS feed and email us when a new item appears — the same real task we tested Make.com against.

Signing up

The signup flow asks for a company size, role, how you heard about n8n, and whether you already know what you want to automate — standard onboarding, nothing that slowed things down. No card, no trial-expiry tricks.

The AI assistant actually built it

We typed one sentence describing the automation into n8n’s assistant panel. It asked one clarifying question (which email service to use — we picked Gmail), then built the entire two-node workflow itself: an RSS trigger watching solo-stack-lab.pages.dev/rss.xml, connected to a Gmail send action.

Before we touched anything manually, the assistant had already run a simulated test of the full chain and told us plainly that it was simulated — “no real email went out during testing.” That’s a small detail, but it matters: Make’s scenario log reported a scenario as “completed” even when a step silently failed. n8n’s assistant volunteered the opposite information without being asked.

We then asked for a real test. It requested one-time permission to execute the workflow, sent an actual email, and reported back the Gmail message ID. We checked the inbox ourselves — the email was there.

The default that needs changing immediately

The assistant had set the RSS trigger to poll every minute. The free plan allows 1,000 executions/month. At one check per minute regardless of whether anything new is found:

60 checks/hour × 24 hours ≈ 1,440 executions/day

That’s the entire monthly free allowance gone in under 17 hours, from a single trigger doing nothing but checking. This is a more aggressive default than Make’s (every 15 minutes, which at least takes about 10 days to exhaust the equivalent free tier).

The fix was trivial: we asked the assistant in plain English to change the interval to every 6 hours, and it did — without re-running the verification, since it correctly recognized this was a minor edit to an already-tested workflow. But the default itself is a trap for anyone who publishes a workflow and walks away without checking it.

OAuth scopes: no repeat of the Make problem

We granted Gmail “view, compose, and send” access up front (informed by the Make review’s scope debugging), and the connection worked on the first real test — no silent failures, no digging through isolated module tests to find a hidden 403 error.

Bottom line

For going from a plain-English description to a working, verified automation, n8n’s assistant was faster and more transparent than manually wiring up Make — it disclosed its own simulated test as simulated, and adjusting a setting afterward didn’t require rebuilding anything. But like Make, the free tier has a real trap in its default trigger frequency, and this one is worse: check and adjust the polling interval before you publish anything, not after.

See our Make vs n8n comparison for the full head-to-head on the exact same task.