Course A · Module A1 Foundations · Lesson 3 of 17
Leave this lesson with one workflow to automate and one tool to build — your north stars for the rest of this course.
Every lesson from here on aims at something real in your business. That means you need a target before we go further. This lesson is hands-on: you'll list your recurring tasks, score them on a quick three-factor formula, and walk out with two picks — one workflow to automate and one small tool to build. Those two picks travel with you through the entire course.
Open a blank document. Give yourself five minutes to list every task you did last week that you've done before — the recurring stuff. Don't filter yet. Aim for at least eight items. Then score each one on three dimensions:
Multiply the three numbers for a score out of 27. The highest scorer is your automation target.
A task that scores 18 or above is a strong candidate. One that scores under 9 probably needs too much human judgment and isn't worth touching yet. That's fine — you're picking the best first step, not automating everything at once.
Separate from automation, think about a small tool you wish existed. Not an enterprise platform — something specific to your business that no off-the-shelf app quite does. Common examples:
Write one sentence: "I wish there was a tool that ____." That sentence becomes your build brief in the later modules.
Before you commit to either pick, run one final question from Lesson A1's Automatable Task Test: what happens if it goes wrong?
You don't need a perfect, risk-free workflow. You need a recoverable one. Almost every first automation qualifies, as long as a person sees the output before it matters.
Think of your first automation as a trial employee. You wouldn't send a brand-new hire to close a deal alone on day one. You'd have them draft the email while you review it. Same idea: keep a human in the loop at first, then reduce oversight as trust builds.
Write these down now — you'll reference them in every subsequent lesson:
Answer before opening. Recalling it now is what makes it stick.
Probably not yet. It scores 6 out of 27 — the low rule-based-ness means there's too much judgment involved for an AI workflow to handle reliably. Look for a task with a higher rule-based-ness score first.
Add a human-in-the-loop checkpoint. Have the automation draft the message and drop it into a "needs approval" queue rather than sending automatically. You review, then send. Full automation can come once you trust the output.
Because finishing one builds more momentum than starting five. Multiple partial projects teach you nothing about the full cycle. One completed automation — even a simple one — shows you the whole loop and gives you something working in days.
Read this next (primary source)
"Building Your First AI Feature" (Mean CEO blog) — practical guidance on starting with one narrow job and keeping a human review step, written for founders rather than engineers. ~8 min. See RESOURCES.md for more.
Key terms from this lesson — automation, human-in-the-loop, trigger, action — are in the course glossary. Keep it open as you go.