Measure Your Baseline

In Week 10 you'll build your own AI system and show what it changed. The way you show that is a before-and-after number.

This lesson is the "before". It takes about twenty-five minutes, it can't be done afterwards, and it's the piece most often missing from weak capstone write-ups.

---

1. Why this can't wait

You might think you can estimate it later. You can't, for a simple reason: people are consistently wrong about how long familiar tasks take. Memory rounds down the boring parts and forgets the interruptions.

Once AI is part of how you do something, you'll never again be able to see accurately how long it took without it. Right now, you still can.

2. Pick the task

You're looking for one task with this shape:

• You do it at least weekly
• It's roughly the same each time
• It takes at least ten minutes
• It involves reading, writing, sorting, summarising or moving information between places

Good examples: the weekly report, clearing a shared inbox, turning meeting notes into actions, preparing for client calls, formatting data for a spreadsheet.

Bad examples: something you do monthly (too few repeats to learn from), something that's mostly a phone call, or something different every time.

It doesn't have to be your capstone. You'll choose that in Week 10. But if you already suspect what you'd like to improve, measure that.

3. Do it and time it

Do the task the way you normally do. Don't change anything, don't try to be faster than usual. Time it with a real timer.

Then record:

Task
Date
Time taken — measured, not estimated
Steps involved
Where the time actually went
Interruptions or waiting
Quality, 1–5, judged by you
How annoying, 1–5

"Where the time actually went" is the most useful row. People usually expect the writing to take longest and discover it was finding the information.

4. Do it twice if you can

One measurement can be an unusual day. If the task comes round again during Week 0, time it again and note both. Use the average.

5. Save it somewhere you'll find

Put it somewhere you'll be able to find ten weeks from now — not a sticky note. A saved document, a note in your usual app, an email to yourself with a clear subject line.

In Week 10 you'll redo the same task with what you've built, measure it the same way, and compare.

6. What a good baseline looks like

Weak:

Weekly report. Takes a while. Annoying.

Strong:

Weekly client report, measured 14 Sept. 52 minutes. Most of the time (31 min) was pulling numbers from three different sheets and reformatting them; the writing itself was about 12 minutes. Two interruptions. Quality 3/5 — usually a formatting mistake. Annoyance 5/5.

The strong version already tells you where AI will help most: not the writing, the gathering. That's the kind of insight a real measurement gives you and an estimate never does.

---

⚠️ Common mistakes

• Estimating instead of timing. The whole point is a real number.
• Doing it faster than usual because you're being watched — even by yourself.
• Picking something you rarely do.
• Not writing down where the time went. That's where the insight is.
• Losing it. Save it somewhere you'll find in ten weeks.

---

That's Week 0. When all five lessons are marked complete, you're ready. Your first live session is on your dashboard's Live schedule tab — see you there.