Why this mattersStalling is a picking problem, not a tool problem
Most owners who stall with AI don't have a tool problem. They have a picking problem — thirty options, no starting line. By the end of this lesson, you'll be able to name the one recurring task that eats the most of your week and say roughly what it costs you in hours. That single decision turns "AI" from a fog into a project.
The explanationAudit, score, pick one
Here's the method: audit your week, score the candidates, pick one.
First, list 3–5 tasks you do more than once a week. Not projects — tasks. "Marketing" is a project. "Writing follow-up emails to slow-paying clients" is a task. If a task has a clear start and a clear finish, it counts. If it drifts, it doesn't — you can't measure a blur, and measuring comes next lesson.
Second, score each candidate with two numbers you already know roughly: how many minutes one run takes, and how many times a week it happens. Multiply. That's your weekly cost in minutes. Estimates are fine here — lesson 2 replaces guesses with real numbers.
Third, pick the task with the biggest weekly cost that still feels boring. Boring is the point. Repetitive tasks have a shape — same inputs, same steps, same output — and that's exactly the shape AI handles best. The exciting, creative work can wait. It's also where AI disappoints first, which is how people conclude AI is overrated and quit.
Where the task lives. It should sit in admin, marketing, or customer service — the areas where these tools genuinely perform.
How it runs. It should be doable in a free browser tool with no new software. If your worst time-sink needs special software or a paid plan, park it. There's a cheaper first win somewhere else on your list.
One task. Not a stack of three. Every extra task in week one makes it less likely you finish any of them.
Worked exampleDana picks onboarding
Dana runs a one-person bookkeeping practice, about 15 clients. She lists her weekly candidates and scores each the same way you just scored yours: minutes per run, times per week, multiplied.
| Task | Min / run | × per week | Weekly cost |
|---|---|---|---|
| Onboarding new clients | ~30 | 5–7 threads | ~3.5 hrs |
| Monthly reconciliations | bursts, full attention | monthly | poor first pick |
| Chasing missing receipts | unpredictable | some weeks | poor first pick |
The numbers say reconciliations are bigger in total — but they arrive in bursts and need her full attention, and receipt-chasing is too unpredictable to measure cleanly. Onboarding wins: repetitive, weekly, self-contained. Roughly 3.5 hours a week — half a workday of email choreography she could do half-asleep. Boring is the point. It's the right first target.
Do it nowTen minutes, tonight
Paper or a blank text file — nothing to install, no new account:
- Write down 3–5 tasks you repeat weekly.
- Score each: rough minutes per run × times per week.
- Circle the highest number that still feels boring. That's your task.
Stop there. Don't start using any tool yet — that's lesson 3, and doing it out of order is how the before/after number gets ruined.
Done when: you've written down one task and its rough weekly cost.
Self-checkThree quick questions
No score, no sign-in — the answers explain themselves.
Q1Which candidate makes the best first workflow?
Right. Repetitive + clear boundaries = measurable, and AI handles shape best.
Not quite — impressive demos impress nobody when the hours don't move.
Tempting, but importance isn't shape. Pick the repetitive one first.
Q2Your weekly cost estimate is which two numbers multiplied?
Right. Minutes per run × runs per week = weekly cost in minutes.
Feelings don't multiply into hours. Time per run × frequency does.
Tool count has nothing to do with the cost of the task itself.
Q3You have a task you do daily and a bigger one you do twice a month. Which gives a cleaner first measurement?
Right. Frequency makes the baseline honest — that's lesson 2's whole game.
Big once is hard to measure. Steady and frequent is cleaner.
Measurement is lesson 2 — and the daily task is where it works best.
What this is bad atThe honest limits
The time audit is only as honest as your memory — a guessed "30 minutes" can really be 45, which is why next lesson replaces estimates with an actual count. It also can't tell you whether AI suits the task you picked: a repetitive task involving client judgment, regulated data, or context nobody wrote down will resist automation no matter how many hours it costs. If your top candidate fails those filters, that isn't the method failing — it's the method saving you from a bad first project. Pick the next task down the list.
Baseline it
Estimates drift. Before touching any tool, you'll count today's hours honestly — so the after number means something. One quick registration unlocks the whole course.
Register to continueMark lesson 1 complete above if you've done the audit.
Register — name and email, that's it. Unlocks lessons 2–6.
Continue to lesson 2 and turn your estimate into a real baseline.
