Lesson 001Field guide · AI Workflows & Systems
What Is an AI Workflow?
From one-off prompts to repeatable routines you can rely on.
You'll learn
- The difference between one-off prompting and a repeatable routine
- The four marks of a workflow: trigger, steps, output, review
- Why writing a routine down once keeps paying you back
- What workflows honestly are (and aren't): human-run, not hands-off automation
- How to spot the first task in your week worth systematizing
01Section
The tenth time you solve the same problem
You've used AI enough to have had this experience: you craft a careful prompt, the result is genuinely good, and you move on. A week later the same task comes around — and you improvise a new prompt from memory. It's a little different, a little worse, and you spend ten minutes fixing output you'd already learned how to avoid.
That's one-off prompting: treating every task as brand new, even when it isn't. This path teaches the alternative — treating a repeating task as a problem you solve once, write down, and then just run.
One-off prompting in the wild
Monday: you write a sharp prompt that turns meeting notes into a clean action list. Thursday, same meeting, you type a rougher version from memory — different structure, a missed owner, extra cleanup. The task was identical. Only your prompt wasn't.
The shift this path teaches
Stop re-solving solved problems. If a task repeats, the prompt — and the steps around it — get written down once and reused.
Checkpoint
02Section
What makes something a workflow
Plain-language definition
AI workflow — a repeatable routine you run with an AI tool: a defined trigger, the same steps in the same order, a consistent output, and a human review before the result counts.
Trigger
What kicks it off — Friday at 4pm, a meeting ending, the inbox hitting twenty unread. If you can't name the trigger, it isn't a routine yet.
Repeatable steps
The same actions in the same order, including the exact prompt you use — not a fresh improvisation each time.
Consistent output
The result lands in the same shape every run: an action list, a set of drafts, a triage summary.
Review
A defined human check before the output gets used. You stay the editor — always.
Checkpoint
03Section
Why writing it down once compounds
Here's the quiet advantage of a written routine: it can improve. When a run disappoints you, you edit the written version — add a rule, fix the format, close the gap. The next run starts from everything you've learned so far. An improvised prompt starts from zero every single time, no matter how many times you've done the task.
A useful analogy
A written workflow is a recipe card. The first time you cook a dish, you experiment. But nobody re-invents a recipe at every dinner — you write down what worked, and from then on you improve the card, not your memory.
Doesn't writing everything down slow you down?
Only once. Writing a routine down costs a few minutes on a task you've already figured out. Every run after that is faster than improvising, and every improvement is permanent. The only routines not worth writing down are the ones that don't actually repeat — and this lesson's last section helps you tell the difference.
Key takeaway
An improvised prompt improves nothing. A written one gets better every time you run it.
Checkpoint
04Section
Workflow is not automation — and that's fine
One piece of honesty before we go further. Everything in this path is a human-run routine: you notice the trigger, you run the steps, you check the output. Nothing happens while you sleep.
Plain-language definition
Automation — software running a task without a human starting or steering it. Connecting tools so steps fire on their own is real — and it's a later frontier, not this path. Workflows come first.
The pitch you've heard
The AI watches your inbox, drafts every reply, and sends them for you while you focus on real work.
What this path builds
Tuesday, 9am: you paste the morning's inbox into your saved triage prompt, read the summary, and answer the three messages that actually matter.
The second one is achievable this week — and nothing goes out without you seeing it.
Anything that ships without human review carries your name on mistakes you never saw. Run your workflows yourself, reliably, before you even think about wiring tools together.
Checkpoint
05Section
Spotting your first workflow
Workflow candidates hide inside repetition. You're looking for a task that fits most of these:
- You've done it three or more times this month
- It follows roughly the same shape every time
- The output goes to the same place or person
- You already use AI for it — but improvise the prompt each time
- You feel a small sigh when it comes around again
Pause and think: Which task came to mind first while you read that list? That's probably your first workflow.
Checkpoint
Prompt exercise
Find the routine hiding in your week
Copy this prompt into ChatGPT, Claude, Gemini, Copilot, or whichever AI tool you have access to — the website doesn't run AI itself. Answer its questions honestly; vague answers get vague workflows.
I want to find one task in my week worth turning into a repeatable AI workflow. Ask me questions one at a time about my typical week — what I do repeatedly, what I already use AI for, and what I quietly put off. After five or six questions, propose the single best candidate and sketch it as a workflow: the trigger, the repeatable steps, what the consistent output should look like, and what I should review before using the result. Keep it practical — this is a routine I will run myself, not automation.
Reflection: Does the sketched workflow match how you actually work? Adjust the trigger and steps until you'd genuinely run it next week — you'll build it out through the rest of this path.
Quick check
3 quick questions — no pressure
There's no pass or fail here. Answer them all, and we'll show you the answers either way.