On September 9, 2026, n8n released a built-in Assistant that turns a plain-language request into a working automation, runs it, fixes what breaks, and leaves you a standard workflow your team owns.
Most automation projects do not fail because the idea was weak. They stall at the translation step. You know exactly what should happen, but not which nodes to use, in what order, configured how, and connected to which accounts. That gap is where workflows quietly die.
n8n Assistant is a built-in AI feature made to handle that translation for you, closing the distance between describing what you want to automate and actually building it.
From Prompt to a Workflow That Actually Runs
You describe the automation in plain language. If your request is ambiguous, the Assistant asks a follow-up question instead of building the wrong thing, which is more than most generators manage.
Then it places nodes on your canvas and wires them together. It asks for credentials only when it reaches a step that genuinely needs them, not before.
Building is the easier half. Once the workflow exists, the Assistant runs it, reads the execution data, and watches for failures. When a node breaks, it diagnoses why, proposes a fix, applies it, and re-runs.
That loop continues until the workflow completes or until the blocker is something only you can fix, like an account that does not exist yet.
This is what separates it from the previous AI Workflow Builder, which generated a workflow once and stopped. The Assistant supersedes that tool. It builds, runs, debugs, and iterates.
What You Keep, What It Costs, and Who Gets It
The output is a standard n8n workflow. It stays visible on the canvas, editable by hand, with full execution history, and it is owned by your team rather than tied to whoever wrote the prompt.
A colleague can open it weeks later, read the nodes, and change a filter or add error handling without knowing what prompt created it.
Usage draws on your plan’s AI credit allocation, counted separately from the old builder. More debugging cycles cost more credits. n8n says it will add ways to top up credits in the coming weeks.
Availability is uneven for now. The feature is on by default for new n8n Cloud instances, with Enterprise Cloud excluded. Self-hosted Docker requires version 2.36 or later, plus your own API keys and extra environment variables. Self-hosted via npm is not supported, and Enterprise self-hosted is on the roadmap.
It ships behind a preview flag, meaning it is in active development with more features coming. Documentation is live at docs.n8n.io/build/ways-of-building-workflows/n8n-assistant.
The real shift is not that a machine can write your workflow. It is that the machine now stays until the workflow works, then hands you something you can still edit by hand.
Do you want to learn how to host n8n by yourself? CLICK HERE.