Automate a local business
Somebody else's business, somebody else's process, and a handover where you never get called again.
What this is
The other Launch capstone. Ship one automation assumes a process you already have access to. This one does not: you have to find the business, earn the access, work with people who did not ask for this, and get out cleanly.
The technical work is the smallest part. Most of the difficulty is that nobody here shares your assumptions, and one of them is going to be the person who has done this job by hand for nine years.
What you need before starting
- The ability to scope and price work. Do the Earn course first, genuinely.
- A real local business willing to talk. Not a friend's hypothetical.
- Eight weeks, most of it waiting for a live system to prove itself.
What you'll leave with
- A discovery write-up for one business, with the task chosen on volume rather than on interest.
- A running system that a non-technical owner operates without you.
- A before-and-after measured the same way, and an honest account of what did not work.
A word on who this affects
You are changing how somebody's staff spend their day. Do that with them in the room.
Module 0 — Find the business
Nobody is waiting for you. This module is about getting into a room you have no reason to be in, and finding out what actually happens there. Free.
- Which businesses have work worth doing — The good ones are boring, profitable, and have somebody staying late on a Thursday. (30 min)
- Find the business and shadow the work — A week of watching somebody do their job beats any amount of asking them about it. (30 min)
- The half hour that decides everything — You get one conversation. Spend it asking, not explaining. (40 min)
- What nobody says out loud — The person who has done this for nine years cannot tell you how. You have to watch. (45 min)
Module 1 — Earn the access
Somebody has to want this, somebody's data is involved, and somebody has to pay. All three are settled before you build anything.
- Your champion, and the person who is threatened — Two people decide whether this survives, and only one of them is on your side by default. (45 min, paid)
- Whose data is this — Customer records belonging to somebody else, handled by you, on your laptop. Get this right. (45 min, paid)
- Price it before you build it — What to charge, how to structure it, and why the first one being cheap is fine but free is not. (60 min, paid)
- The agreement, in writing — Two pages that prevent every argument this kind of work produces. (45 min, paid)
- When to walk away — Some engagements should not happen. Recognising them early is worth more than any technique in this course. (40 min, paid)
Module 2 — Build for the operator
The person running this will never open a terminal, will never read a log, and will not tell you when it is wrong — they will quietly stop using it.
- Build for somebody who will never read a log — The operator is not you. They need to know when it is wrong, not why. (35 min, paid)
- One screen, no options — The interface a busy person uses at 8am with a customer waiting. (45 min, paid)
- The wrongness signal, designed — A non-technical operator has to be able to tell, in the moment, that an output is bad. (45 min, paid)
- The exception path in their words — Where the awkward ones go, described so somebody who did not build it knows what to do. (40 min, paid)
- The software they already have — You are not arriving at an empty desk. Something already runs this business, and somebody sold it to them. (45 min, paid)
Module 3 — Inside their business
Running it where the stakes are somebody else's. The technical part is smallest; the weekly conversation with your champion is what decides it.
- The weekly session with your champion — An hour a week going through disagreements together, which is where the requirements actually come from. (60 min, paid)
- The operator drives — Switch a quarter of the volume, and let somebody who is not you run it. (45 min, paid)
- The week you are not there — Go away for five days. What happens is the real status report. (45 min, paid)
- What it costs them to run — Their bill, their queue time, and the honest hours-a-week figure. (45 min, paid)
- When the pilot does not work — Sometimes the honest answer is that it did not help. Delivering that well is a skill. (45 min, paid)
Module 4 — Leave properly
The measurement, the handover, and the part where you stop being needed — including the commercial conversation nobody has.
- Leave properly — The measurement, the handover, and the part where you stop being needed. (30 min, paid)
- The support conversation nobody has — Price the ongoing relationship, or end it. The middle option is eighteen months of unpaid favours. (45 min, paid)
- The case study, anonymised — The method is the portfolio piece. Their name is optional and their numbers are not yours to publish. (60 min, paid)
- The next three jobs — The second engagement is where this becomes a business rather than a project. (45 min, paid)
Related writing
- How to choose the first thing worth automating — Most first automations fail on selection, not engineering. Four tests that separate a task that will survive contact with real use from one that quietly gets abandoned.