The automation mindset
Every automation is a trigger, steps, and an output. Learn the anatomy, meet the two kinds of steps, and see the week of work you're about to get back.
Meet Riley, operations coordinator at Harbor Lane, a specialty coffee retailer with a web store and three physical locations. Riley's Tuesday: copy order-issue emails into the ticket tracker (40 minutes), assemble the weekly ops report from four browser tabs (2 hours), forward review alerts to the right store manager (whenever they remember), chase unpaid wholesale invoices (dreaded, deferred). None of this is Riley's job — it's the connective tissue around the job. This course is about handing the connective tissue to software.
The anatomy: trigger → steps → output
- A trigger starts the run: an event ('new email arrives', 'row added to sheet', 'form submitted'), a schedule ('every Monday 7am'), or a manual button. One trigger per workflow.
- Steps do the work, in order: read this, look that up, transform it, send it there. Each step's output feeds the next step's input — a bucket brigade for data.
- An output is the point: a ticket created, a message sent, a report drafted, a row updated. If you can't name the output, you don't have a workflow — you have an aspiration.
Everything you'll build — in any tool, at any scale — is this shape. Riley's email-to-ticket chore: trigger (email arrives at support@) → steps (extract sender, subject, order number; add row to tracker; post to the ops channel) → output (a logged, visible ticket, in four seconds, at 2am, without Riley).
Two kinds of steps — and the course in one sentence
- Deterministic steps follow rules exactly: copy this field, add a row, send at 9am. Same input, same result, every time. Cheap, instant, boringly reliable. Most of every workflow should be deterministic.
- AI steps exercise judgment on messy input: 'what is this email about?', 'how urgent does this customer sound?', 'draft a polite reply.' Powerful on exactly the tasks rules can't touch — and probabilistic, meaning usually right. AI steps get used deliberately and checked (Modules 3 and 4 are entirely about this).
The course in one sentence: use deterministic steps for everything rules can handle, add AI steps only where judgment is genuinely needed, and wrap anything consequential in human checkpoints. Learners who internalize that sentence build automations that run for years. People who skip it build demos that get quietly turned off.
The big low-code platforms — Zapier, Make, n8n, Power Automate — differ in pricing and polish, not concepts. Triggers, actions, paths, and AI steps exist in all of them under slightly different names. This course teaches the concepts with platform-neutral instructions; labs include a 'translate to your tool' step. Use whatever your workplace already has — the skills transfer whole.
No prerequisites, no programming. If you can write a clear email and follow a recipe, you're qualified. The one skill you'll build that feels like programming is thinking in steps — and you already do that every time you explain a process to a new coworker.