From the desk
Before The Admin Desk, we ran the office of a busy residential building business: quotes, claims, timesheets, payroll, suppliers, marketing and a constantly ringing phone. This is what changed when we put AI to work, and what we learned about checking it.
A small residential builder in South Australia: a licensed builder on the tools, a site supervisor, a qualified carpenter and apprentices, and one person running the entire office. Renovations, extensions, bathrooms, decks and verandahs, with several jobs running at once.
The software was already good: job management and estimating in Wunderbuild, accounts in Xero, everything in Microsoft 365. The problem wasn't the tools. It was the hours spent moving information between them, checking them against each other, and writing the same things over and over.
Before: opening every job, one by one, to see what had been done, what had been claimed and who owed what.
After: AI connected to Wunderbuild and Xero reviews the whole portfolio in minutes. In its first review it found a backlog of draft claims that had never been sent, which became the single biggest cash flow fix. It has since flagged jobs claimed ahead of their stage, and invoices that never synced to Xero.
Before: working out labour for every quote from scratch, with the risk of forgetting something.
After: two small custom tools built with AI. A labour quoting assistant and a quick job estimator with presets for each job type use the business's own rates and margins, so a labour price takes minutes and is the same every time.
Before: an inbox of nearly 4,000 emails, timesheets checked by hand each fortnight, and BAS prep that took most of a day.
After: a folder structure built around how the business actually works, a digital to-do system on the phone, AI timesheet checks before every pay run, and BAS prep that cross-checks job receipts against Xero automatically.
Before: social media was the job that always got pushed to Sunday night, or didn't happen.
After: a 6-week content plan, captions and carousel copy in an afternoon. Job photos become posts in minutes, in the business's own voice.
These are our own estimates of time saved in an average week, compared with doing the same work by hand.
| Area | Time saved per week |
|---|---|
| Claims, cash flow and Wunderbuild/Xero checks | About 1.5 to 2 hours |
| Labour pricing for quotes | About 2.5 to 3 hours |
| Inbox, timesheets and BAS prep | About 2 to 2.5 hours |
| Marketing and social media | About 2 hours |
| Total | About 8 to 10 hours, more than a full working day |
Over a year, that's more than 400 hours. More importantly, much of it used to happen at night and on weekends.
AI didn't just save time. It found problems a person would likely have missed: unsent claims, claims ahead of the work, and invoices that never made it from one system to the other. But it also showed why a person always has to check. The numbers are only as good as the data, systems don't always sync the way you expect, and AI will confidently fill a gap with a guess.
So every workflow we set up has the same rule: AI does the gathering, sorting and first draft. A person checks it before it goes to a client, the accountant or the bank.
Your business next
Every business has its own version of this. We'll look at how yours runs, find the double-handling, and show you what AI could take off your plate, with a human double-check built in.
