Your first-automations workbook.

A short working session for you and your team, about 30 minutes. Don't just read it, fill it in. By the end you'll have your first flow picked and planned, a build checklist ticked off, your testing and error handling sorted, and a 90-day plan. Type straight into the boxes and save as a PDF, or print it and write on it.

The model, in brief. An automation is "when this happens, do these things in order." The trigger is the when, the actions are the things, and data flows down the chain so each step uses what came before. Build with Make or Zapier, add an OpenAI or Claude step to summarise, classify or draft, and keep a human check on anything customer-facing. Test one record before you trust it. Graduate to n8n when cost or complexity outgrow the no-code tools.

1What to automate first

Brain-dump the repetitive jobs your team would happily hand off, then score each one. A good first job is frequent, repetitive, resented, and low-risk if it's slightly off. Quantity first, then pick.

The repetitive job	How often	Pain (1-5)	Risk if off

The one flow to build first:

2Trigger, action and AI-step planner

Plan your chosen flow as "when X happens, do A then B then C." Name the trigger, list the actions in order, and decide what the AI step does (if any) and whether its output needs a human check.

Trigger (the "when": a form, an email, a schedule): Actions, in order (save to CRM, send email, add a row, alert someone): AI step (summarise, classify, draft) and the output format you need: This AI output reaches a customer or carries risk, so a person approves it before it goes out. Apps I'll need to connect:

3Build checklist for your first flow

Work through these as you build in Make or Zapier. Tick each one off and you'll know the flow is wired right before you switch it on.

I sketched the flow as one plain sentence before opening the tool. Every app is connected, using the account I want it acting as. Each field from the trigger is mapped into the right box on every action. Required fields and formats are handled (a filter for missing essentials like email). Where AI output is risky, the flow drafts it and routes it to a person to approve.

4Reliability and testing checklist

This is what turns a clever demo into something you can depend on. Test before you trust it, and build for the day it breaks.

I ran one test record through and followed it into every app. I tested the awkward cases: blank field, huge message, obvious spam. Failing steps retry, so a momentary outage doesn't drop a lead. Failure notifications are on, so a break becomes a message I can act on. I'll watch the run history daily for the first week or two. Rough cost to watch (Make operations / Zapier tasks / AI per-run):

5Your 90-day plan

Keep it small and real. Ship one reliable flow, prove it against the old manual way, then add the next. Three solid automations beat ten half-finished ones.

Flow 1 (ship first, with error handling and a week of watching): Flow 2 (once flow 1 is earning trust): Flow 3 (the next win): How I'll measure each one against the old way (time saved, leads caught): When might I outgrow no-code (cost, complexity, control) and need n8n or a hand:

Finished it?

Do something with it.

Use Save as PDF up top to keep a copy for your team. Or send it to me for a free second opinion, and I'll email a copy of your answers to your inbox as well.

Your first-automations workbook.

Sent, thanks.