TL;DR - Using AI manually means you prompt and it responds. Automation means systems run without you. The skill is judging what to automate - frequent, rule-based, low-judgment tasks - and what to leave alone.
Why it matters
Many people automate too early or too much, then spend more time fixing brittle workflows than they saved. Good judgment about candidates is the whole game.
Worked example - sort your tasks
"Log every contact-form submission to a sheet and Slack me." -> automate (frequent, rule-based, clear trigger).
"Draft a tricky reply to an upset customer." -> use AI manually (needs your judgment).
"Decide our pricing." -> keep fully human (high-stakes, rare).
Steal this - the automation test
Automate a task only if it's mostly YES on:
1. Do I do it often? (daily/weekly)
2. Same steps each time?
3. Clear trigger (an email, a form, a schedule)?
4. Low judgment?
The AI pattern: trigger -> AI understands/transforms -> action.Common mistakes (and the fix)
- Automating rare or judgment-heavy tasks. Fix: keep those manual.
- Boiling the ocean. Fix: automate one annoying step reliably first.
- No monitoring. Fix: a silent failure can be worse than doing it by hand (covered later).
Good to know
Automation lives in Zapier, Make, and n8n (next lessons). Plain automation just moves data; adding an AI step (classify, summarize, draft) is what makes it intelligent - the theme of this level.