The Hours Nobody Clocks

Ask an owner what a job costs and you'll get an answer to the dollar — materials, labor, drive time, margin. Ask where Tuesday afternoon went and you'll get a shrug.

It went where it always goes. Retyping the overnight requests into the scheduling sheet. Writing the same "can you give me a ballpark?" reply for the fourth time today. Pulling hours, mileage, and receipts into the Friday summary. None of it shows up on an invoice. None of it gets timed. It's just how the week is.

You price every job you do for a customer. The jobs you do for yourself — twenty times a week — never get priced at all.

The stopwatch on the shop floor

There's a century-old fix for this. Factory managers used to walk the floor with a stopwatch and a clipboard and time everything — every reach, every walk to the parts bin, every form filled in twice. Not because the crew was slow, but because nobody can feel where minutes leak. The waste hides inside "how it's always done." Once it was on the clipboard, it was obvious.

Nobody walks their own office with a stopwatch. So the repetition never makes it onto a clipboard, and "I should really automate some of this" stays a feeling instead of a plan.

Why "use AI more" goes nowhere

Most AI advice tells you to use it more without ever asking where. The numbers show how that ends: around 78% of companies now use generative AI, yet roughly 80% report no significant bottom-line impact (McKinsey research, cited in AI READY by Danny de Vries). Everyone's using it; almost nobody can point at the hour it gave back.

The failure usually isn't the tools — it's the aim. Only about 21% of companies actually rethink how the work flows when they bring AI in; the rest bolt it onto whatever they were already doing (AI READY). Small businesses make the same mistake at small-business scale: automate the flashy thing, skip the boring one that's quietly eating an hour a day.

The question was never "should I use AI?" It's "which hour do I hand it first?"

A review you can run before lunch

So I built the audit nobody does. The Busywork Review is a page where you describe your week in plain sentences — the tasks you do on repeat, in your own words, mess and all. It scores each one against a simple fit test called RETS: is the task Repeating, Rules-based, Text-shaped, and Safe-to-draft? A task that passes all four is where AI earns its keep first. Then it hands you one ranked answer — not twenty ideas. The single biggest time-sink that's actually safe to hand off, and a plain-English step you can take this week in under an hour.

Just as useful: it tells you what should stay human. Driving out to look at a property before you quote it fails the test on purpose — that's in-person judgment, the part customers are paying you for. A review that only ever says "automate everything" isn't a review; it's a pitch.

And if you'd rather keep the review than visit it, the same thing exists as an installable skill — a plain-English instruction file you paste into whatever AI assistant you already use, then re-run any time your week changes.

What it touches

Plainly: the description you paste is sent once to an AI model to score it, and the review shows up on the page. Nothing is stored — no account, no database, no list. Close the tab and it's gone.

And what comes back is a draft you confirm. It recommends what to hand off; it never tells you to remove a person from anything that reaches a customer. Any time figure it wasn't told is labeled an estimate, never presented as fact. The tool reads; you decide.

Do the five-minute audit

I've shipped north of 20,000 software products, and the pattern never changes: the wins come from aiming at one boring, specific task, not from buying more capability. Write down the stuff you do on repeat, exactly the way you'd complain about it to a friend. Then paste it in and see what the clipboard says.