Internal build
Bus Parking Wiki: one place for what a driver needs at the gate.
A phone app I built for bus and coach drivers in Victoria. It went live behind a sign-in the evening its plan was finished. Here’s what it holds after one week, the rules it works to, and what it hasn’t proven yet.
The last few hundred metres.
A bus GPS gets you to the street. It doesn’t tell you which gate the school wants buses to use, whether there’s room to turn around, or where you can wait for a few hours without breaking a road rule.
Some of that is published, tucked into an excursion note for teachers or a council parking page. Most of it lives in experienced drivers’ heads, phones and notebooks, and it leaves with them when they retire. At the 2021 Census, more than half of Australia’s bus and coach drivers were 55 or older (Jobs and Skills Australia).
I run Lumomatics, and I also drive charter coaches, so I’ve met this problem first-hand.
The same problem in your business.
Swap the coach for a ute and the school gate for a client’s side gate. Which clients need a call before you arrive, which supplier only answers before 7am, how to handle a cancellation when the usual receptionist is away. It usually lives with one person, and nobody else can get at it when they’re sick, on leave or gone.
Finding that kind of knowledge, and working out the simplest way to get it out of one person’s head, is what a Workflow Clarity Audit is for.
What it does.
- One card per site, in the order a driver meets it: approach, entry, drop-off and pick-up, waiting, exit and rules.
- A Directions button that opens the driver’s own navigation app at the card’s bus entry or drop-off pin when it has one, not just the street address.
- Low bridges near the site, checked against the height of the driver’s own bus.
- Cards saved on the phone still open with no signal.
- Each card shows whether a driver has checked it on site, and if so, who and when.
- Drivers can report a problem, add a photo or add a missing place from their phone, and I get an email for each one.
How it was built.
The plan was finished on a Saturday afternoon and the app was live behind a sign-in that evening. In the first seven days, 40 changes went live: low bridges, school camps, accommodation, wineries, rest areas, coach parking, rail replacement stops, pools, racecourses, photo privacy, backups and email alerts.
| At the end of the first week | How many |
|---|---|
| Places listed | 6,382 |
| Written cards | 1,094, with 2,179 links back to official sources |
| Low bridges (5.0 m or lower), reloaded nightly | 672 |
| Highway rest areas and truck bays, reloaded nightly | 457 |
| Public toilets | 6,088 |
| Cards checked on site by a driver | 3 so far |
Claude, Anthropic’s AI, did most of the coding. My part was deciding what the app is for, agreeing the plan and the rules, and deciding what went live. Each change came to me with a short plain-English description, and I merged it from my phone.
The rules sit in a file the AI reads before it changes anything. A few of them:
- Never round a bridge clearance up.
- No student names, rosters or photos of children.
- Nothing copied or traced from Google.
- Never invent a pin, a clearance or a bus bay. Unknown stays unknown.
Writing a rule down doesn’t guarantee it’s followed. The bridge rule went in on day 2, but older code still showed a card’s lowest clearance to one decimal place, so 3.86 m appeared as 3.9 m. For a bus 3.88 m tall, one figure says stop and the other says go. It was fixed on day 6, and an automated test now checks it. Deciding which rules matter, and checking that they actually hold, is the part I’d spend the most time on with a client.
The parts nobody sees.
These are the easiest parts to skip when something’s built quickly. I spent 25 years in IT support and infrastructure, including health services, and they’re what I’d check first on any system I was responsible for.
- Privacy. Photos lose their GPS location, time and phone details on the server before they’re stored. School cards hold vehicle logistics only: no student names, rosters, pick-up times or photos of children.
- Security. Drivers sign in with a one-time code sent by email, and only drivers I’ve approved can open the app. The app checks the sign-in itself and refuses to open without it.
- Backups. A weekly backup copies the database away from the system it runs on and test-restores it; the first run passed. A Mac mini is set to copy the database and photos to Dropbox every four weeks.
- Honest data. Official information is rewritten and linked, never copied, and licence conditions are followed. If a gate’s location isn’t known, there’s no pin rather than a guess.
- Checks. Before the AI sends me a change, the rules have it run eight sets of automated checks, including one that clicks through the app in a real browser.
What it hasn’t proven yet.
A working app isn’t the same as an app drivers rely on. The trial with a small group of invited drivers has just started, and three cards have been checked on site so far.
Over the next 90 days I’m watching four things: whether drivers use it before trips (I’ll ask them, because the app deliberately doesn’t track which cards they open), whether they send 10 or more corrections and tips, whether 50 places get a driver-checked date, and whether at least five cards come from someone other than me. I’ll update this page with the results.
Questions about this build.
Is Bus Parking Wiki a Lumomatics client project?
No. It’s my own app, built with the same tools I use for client work.
Can I see it?
The public start page is at busparking.com.au, and there’s a six-minute video of the app on YouTube. The app itself is for approved drivers, but I’m happy to show it to you in person.
Would my business need a custom app like this?
Probably not. I start with the problem and the day it shows up in. Often the fix is a change to tools you already have.
How long would something like this take for my business?
It depends on the problem. This one moved quickly because I was the builder and the user, and I already knew the work. With a client, the audit comes first, so we agree what’s worth building before anything is built.
Your business
Where to start in your business.
A Workflow Clarity Audit starts by walking through one recent day with you. We find where the knowledge is stuck and what it’s costing, and I recommend the smallest fix that will work. Sometimes that’s an app. Often it’s a shared checklist, better use of software you already pay for, or an AI receptionist that answers the questions you get every day. The audit is AUD $399, fixed scope.