Once you can build one automation, the way to get fast is to stop starting from scratch, because most automations are variations on a small number of shapes. This lesson gives you those shapes so you have a starting point for almost anything.

The most common shape is trigger, do, notify: something happens, the flow does the work, and it tells a human it is done. A form comes in, the details get saved, and you get a WhatsApp message. Simple, and enormously useful, it is probably half of all the automation you will ever want.

The next step up is enrich with an LLM: somewhere in the middle of the flow, you add an AI step that reads something and produces something, classifying an incoming message, summarising a document, drafting a reply, and the rest of the flow carries that result onward. This is where "automation" and "AI" meet: the flow is still a predefined workflow, but one of its steps is an LLM doing a judgment task a fixed rule could not.

The third shape worth knowing is human in the loop: the flow does most of the work but pauses for a person to approve before it does something consequential. When an automation can send an email to a customer or spend money, you often want a human check before the irreversible step, and building that pause in deliberately is a mark of someone who has automated real things and been burned by fully-automatic mistakes.

The practical habit is to browse n8n's template library, find something close to what you want, and adapt it rather than building from nothing. Recognising which of these shapes a task fits is what lets you go from "I could automate that" to "that is a trigger-do-notify, I will have it in ten minutes."