ai_tribe Free exercise
Where your time actually goes
You signed up to build an assistant. You will not build one in this module, and that is deliberate.
Almost every one of these projects that fails, fails right here, before anything technical happens. Somebody hands over a job that nobody had looked at properly, and six weeks later they have an expensive way of doing something that did not need doing at all.
The work in this module is not exciting. It is a week of writing single lines on a page. It is also the only part of this whole thing that nobody can do for you, because it is your week and you are the only one who sees it.
By the end you will have five jobs written down, ranked, with the first one named. Everything we build after this sits on that list.
Most people skip this module and come back to it. It is very hard to hand over work you have never looked at properly.
What this takes
Ten minutes now to read the rest of this page. Then a few seconds at a time across five working days. Then about half an hour at the end of the week to sort what you collected.
Pen and paper is completely fine. If you would rather print something, the worksheets are here.
Part one: log your week
For five working days, every time you finish a task, write one line. Not at the end of the day, when you have already forgotten half of it. When you finish it.
Three things per line, and nothing else:
- What it was. Six words is plenty. “Emailed Karen the driveway quote.”
- Roughly how long. Round to the nearest five minutes. Nobody is auditing you.
- Would you do it again next week? Yes or no.
Four rules that make the log worth having
Log the small things. The two minute jobs are the ones that hide. Twelve of them a week is an afternoon a month, and not one of them ever felt worth writing down at the time.
Do not tidy your week. The temptation is to have a productive week because you are watching yourself. The messy week is the real week and the messy week is the one worth fixing.
Do not analyse while you log. You will start noticing patterns on day two and you will want to start solving them. Write it down somewhere else and keep logging. Analysis during collection quietly changes what you collect.
Include the interruptions. The phone call that took eleven minutes and broke your afternoon is data. So is the third time this week you answered the same question about pricing.
If five days is not going to happen, do three. Three days is enough to see the pattern. One day is not, because one day is never a normal day.
Part two: sort into three columns
At the end of the week you will have somewhere between forty and a hundred and fifty lines. Now sort every one of them into one of three columns.
Repeatable. Same steps, same shape, every time. Someone else could do it if they knew what you know.
Judgment. You have to decide something, and the deciding is the actual work. Whether to take the job. Whether this client gets the discount. Whether that price is still right.
Relationship. It mattered that it was you. The call where they were upset. The site visit where you read the room.
Only the first column is a candidate. Not because the other two are unimportant, but because they are the opposite. They are the reason the business is yours, and they are what you get back when the first column stops eating your week.
Most jobs are mixed, and this is where people go wrong
Quoting looks like judgment, so it goes in the judgment column and stays there forever.
Look closer. Pulling the current prices, working out the measurements, formatting it the way you always format it, writing the same three paragraphs of terms, sending it, and chasing it on Thursday when they have not replied. That is repeatable. Deciding what to charge this particular client is judgment, and it is about ninety seconds of the hour.
Split the mixed ones. Put the repeatable part in column one and leave the ninety seconds where it belongs. If a job feels like it cannot be handed over, it is usually a mixed job that has not been split yet.
Part three: score column one
Go through the repeatable column and give each job four numbers.
How often. Times per week. Count it, do not estimate it, the log is right there.
How long. Minutes, each time.
Minutes per week. The first number times the second. This is the one that does the convincing, and it is almost always larger than people expect.
How mechanical, one to five. Five means a new person could do it correctly from written instructions and nothing else. One means it needs things that only live in your head.
How bad if it goes wrong, one to five. Five means it costs real money, or a client sees it, or you cannot take it back. One means you would notice, fix it, and forget about it.
A worked example
| The job | Times/week | Mins each | Mins/week | Mechanical | Damage |
|---|---|---|---|---|---|
| Answering the same pricing question by email | 12 | 6 | 72 | 5 | 2 |
| Chasing overdue invoices | 5 | 10 | 50 | 4 | 4 |
| Writing up job notes after a site visit | 8 | 12 | 96 | 3 | 2 |
| Deciding which jobs to quote next week | 1 | 40 | 40 | 1 | 5 |
The last row is judgment wearing a costume. It scored a one for mechanical, which is the tell. It should never have been in column one and it is not a candidate.
Of the remaining three, the pricing emails are the strongest first job: high minutes, completely mechanical, and almost nothing goes wrong if it gets one slightly wrong.
Invoice chasing is a bigger prize than it looks, but chasing the wrong client is a real problem, so it is a four for damage. That does not disqualify it. It means it is a job that needs a boundary around it, and boundaries are what the kit’s constitution module is for.
High minutes, high mechanical, low damage. That combination is your first job. Not the most annoying one, and not the most expensive one. The one you can get wrong safely while you are still learning.
Part four: pick five, rank them, name the first
Take the five highest scorers and write them out properly, in order.
Write each one as a sentence a stranger could understand. Not “quotes”. Try “turning a set of measurements into the quote email I always send, and following it up on Thursday”. If you cannot write the sentence, you do not understand the job well enough to hand it over yet, and that is useful to find out now rather than later.
Then add up the minutes per week across all five, and multiply by forty six working weeks.
That number is what this is worth to you. Write it at the bottom of the page.
One last thing, and it is the useful bit
When you registered, we asked you which job you would hand over first. You answered in about four seconds.
Go and compare that answer with what came out of the audit.
If they match, good, you already knew your week and this confirmed it. If they do not match, that gap is the most interesting thing you will find in this module. In our experience the four second answer is usually the most annoying job, and the audit finds the most expensive one. They are rarely the same job, and people almost always automate the annoying one first and wonder why it did not give them their week back.
What you have now
A ranked list of five repeatable jobs, in your own words, with real numbers attached and one named as first. On paper, from your actual week, not from a template.
That list is worth having on its own. It is a hiring brief, a delegation list, and a decent argument for changing how you do a couple of things regardless of whether you ever automate any of them.
It is also the first thing the kit asks for.
Next: the kit
The kit builds the brain: a folder Claude reads, rules it obeys, a memory that keeps itself current. Module 2 of the kit is the shoebox, where you drop in what the business has already written, and this list goes in first. Module 7 is where you ask the brain about job number one and see the difference between an answer from your business and an answer from a blank chat window. That difference is the whole point.
The kit is $149, one-off, self-paced, on your own Mac. If you come to Build Camp later, the price comes off your seat.