By skill

Plan, Prompt and Fix Automations with ChatGPT: Before and After You Build in n8n, Make or Zapier

What you’ll be able to do: After this course you can hand ChatGPT a plain description of a manual process and get every step sorted into rules a computer can follow and judgments a person keeps, with the questions to answer before anyone builds. You can get a ready-to-paste prompt for an AI step that returns clean JSON, test it on sample emails and tighten it where the test shows a weakness. And when an automation fails, you can get each error explained in plain words, separated into what is certain and what is a guess, then a safe order to fix and test it.

Who it’s for
Office managers, operations leads, assistants and small-business owners who are about to build a first automation or have inherited one. Useful before any of the n8n, Make or Zapier courses. No coding.
Time needed
About 27 minutes of hands-on practice: 3 modules of 9 minutes each.
Cost
Free. No credit card, no trial, and no sign-up needed to start.
Level
Beginner
Prerequisites
None for the practice: a web browser on a desktop or laptop. To repeat the modules in your own tools: a ChatGPT account.
Format
Self-paced and online. Each module runs in a practice copy of the real tool that checks every step, then a checklist walks you through doing it for real.
Tools you practice in
ChatGPT.
3 modulesBeginner~27 min of practiceFree

Why this course

Automation tools are good at running steps. They do not tell you which steps to automate, what the AI step should be told, or why Tuesday's run failed. Those three gaps are where first automations go wrong: the wrong process gets automated, the AI step returns something the next step cannot read, and the first error message stops everyone.

This course fills those gaps with ChatGPT, on three made-up offices: a staffing agency chasing timesheets, an insurance agency sorting a shared inbox, and a small hotel whose booking automation broke after someone renamed a spreadsheet tab. Every answer you see is from a real ChatGPT run on 16 November 2026, trimmed but not reworded. Nothing was built or run inside n8n, Make or Zapier for these lessons, and each module says what to test in your own tool. The tool courses take it from there.

What you’ll build

  • Module 1: You work at Northgate Staffing (made up), where the office manager spends every Friday chasing 60 timesheets and 14 supervisor approvals. You attach her own description and get a 15-step table, each step marked Rule-based, Mixed or Human, and the six questions to answer first.
  • Module 2: You work at Calloway Insurance Agency (made up), where one person reads about 90 emails a day and forwards each to Claims, Quotes, Certificates, Billing or Other. You attach the desks and six sample emails and get a complete, paste-ready prompt that returns five JSON fields.
  • Module 3: You run the front desk at The Alder House, a 22-room hotel (made up). You attach what the automation does, what changed last week and the failed runs, and get each error explained with a "sure" list and a "guessing" list, plus whether Thursday's quick fix helped.
  1. Module 1 · ChatGPT · Beginner · ~9 min

    ChatGPT: Turn a Manual Process into an Automation Plan Before You Build

    You’ll be able to: Turn an office manager's Friday timesheet chase into a 15-step table of rules and judgments, answer the open questions, and get a plan with a trigger, an approval point and a "must never do" line for each automation.

  2. Module 2 · ChatGPT · Beginner · ~9 min

    ChatGPT: Write and Test the Prompt for an AI Step in n8n, Make or Zapier

    You’ll be able to: Get a paste-ready prompt that routes a shared inbox to five desks as JSON, run it on six sample emails, and fix the two weaknesses the test shows.

  3. Module 3 · ChatGPT · Beginner · ~9 min

    ChatGPT: Read an Automation Error and Fix a Workflow You Didn't Build

    You’ll be able to: Read "Unable to parse range" and "Invalid To header" in a failed-run log, see what is sure and what is a guess, then follow a safe order to fix, test and catch up on the four missed inquiries.

Prerequisites

  • For the practice: a web browser. Each module runs in a copy of the real tool’s screens, so you need no account and no sign-up to start module 1.
  • To do it for real: a ChatGPT account. Each module says which one it uses.

Time to complete

About 27 minutes of hands-on practice across 3 modules (60 steps in all). Modules take about 9 minutes each, so you can do one in a sitting and come back for the next.

Questions about this course

Can ChatGPT help me plan an automation?

Yes. Give it the process in your own words and ask it to list every step, what starts it, and whether it is a rule or a judgment, before it designs anything. In module 1 it turned one page of notes into a 15-step table and 34 questions, then a plan once four of them were answered. You still decide what to build first.

What should I automate first?

The step that repeats most, follows a clear rule and can do little harm if it goes wrong, such as a reminder email. In module 1 ChatGPT proposed seven small workflows; the lesson has you start with one, the Friday 9 AM reminder, and add the next only after it runs cleanly.

How do I write the prompt for an AI step in n8n, Make or Zapier?

Tell it the one job, the exact choices it may pick from, the fields to return as JSON, what to do when two choices fit, and what it must never invent. Then test it on real examples before it goes live. Module 2 gives you a full prompt to copy and shows the test.

What does "Unable to parse range" mean in an automation?

It comes from Google Sheets and usually means the step is pointing at a tab or range that no longer exists, often because someone renamed the tab. Module 3 works through that message and Gmail's "Invalid To header", which usually means the email address was empty or malformed.

Is this course free, and do I need an n8n, Make or Zapier account?

It is free: 3 hands-on modules, about 27 minutes in all, in a practice copy of ChatGPT in your browser. You need no automation account for the course. You will want one afterwards to build the plan.

Did you test these automations in a real tool?

No. These three lessons are about the thinking around an automation, and the ChatGPT answers are real, but nothing was built in n8n, Make or Zapier for them. The n8n, Make and Zapier courses on this site are the ones built on those tools' own screens.

← All courses