Justin Rigby was running a national inspection team off an Excel tracker, and running it well. The spreadsheet was accurate. The jobs got covered. Nothing was on fire.

It just took a person, by hand, every week. That's the part we went after.

Justin owns Remedy Asset Protection out of Melbourne. 

His actual job is telling people whether a bridge, a pipeline or a water tank is quietly rusting itself to death. He's flown around the country to do it. The less glamorous half of the week is keeping track of who's where.

Remedy runs inspectors across the whole country. Who's on site at a project in WA. Who's driving to a project in New South Wales. Who's on leave next Tuesday.

All of it starts in a piece of job software called Deltek, exported into five separate saved searches, one per state, then pulled together by hand into the Excel tracker. The data was solid. It just lived in a format that took real time to maintain and didn't show you the whole operation at a glance.

So we spent a couple of weeks building something that does. A live dashboard that turns those Deltek exports into one national schedule you can read in a second, and updates itself when the jobs change. We called it Ops Command.

You export everything out of Deltek in one go. All states, one report. You drop it into Ops Command.

Here's where the time actually went before. That Deltek export comes out as a raw CSV. To make the tracker usable, someone had to manually reformat that CSV in Excel every time, shaping it by hand into the layout the team worked from. That reformatting was a big chunk of the weekly job. Ops Command takes the raw export as-is and does that step for you.

From there it reads the jobs, works out who's scheduled where, and draws the whole national operation as a live calendar and a Gantt chart. Inspectors down one side, dates across the top. The thing nobody could picture before is suddenly just there on one screen.

It flags conflicts on its own. Two activities on the same person, same day, it catches the clash for you instead of someone having to eyeball the spreadsheet for it.

It handles the real-life stuff Deltek never tracks. Leave, training, travel days, public holidays. The things that quietly wreck a schedule but never make it into the job system.

And it stays current. Re-upload a fresh Deltek export and it just updates itself, logging every change as it goes. First proper test run picked up 6 new jobs and 11 changes from a single Victoria upload. No manual cross-checking.

Here's the honest part.

The first real demo was rough. We sat down for a testing session and things broke live, in front of the client. It was easily the most useful hour of the whole build.

Jobs that had their dates cleared in Deltek were still sitting on the calendar like nothing happened. Seven jobs threw a "no state" error even though the state field was clearly filled in. A railway shutdown job had two inspectors working different shifts, and the system put the wrong guy down as lead.

And the thing Justin's team does constantly, jobs that run Monday to Friday, needed a separate entry for every single day. Painful.

None of that goes in a brochure. But that session is the entire reason the tool is any good now.

We fixed the status overrides so Deltek changes actually flow through. Added the manual entries. Killed the "no state" bug. Switched everything to a 24-hour clock, because that's how inspectors read time. Filtered out jobs that have moved into the reporting and drafting phase so the schedule only shows live work.

You don't get to the good version without the ugly demo. Anyone who shows you a clean one first time is showing you a sales deck, not a build.

Then this landed

A couple of things I want to remember.

The data was already there, and it was already good. The win wasn't some clever model. It was taking accurate data the client already owned and giving back the hours it took to maintain it. Most of the "AI in business" wins I keep running into look exactly like this. Not a robot replacing anyone. Just handing back the time a careful person was spending on upkeep, so they can spend it on the work only they can do.

And the rollout pattern in operations is the same one I keep seeing in finance. AI does the obvious 70%. The human keeps the 30% that needs judgement. A senior inspector should be assessing corrosion, not playing human scheduling engine.

If you're running a team off an export and a spreadsheet that one person babysits, you already have the data. You're just doing the reconciling by hand. That problem is more fixable than it looks.

Keep an eye out.

Daniel