Across 35 infrastructure sites and hundreds of subcontractors, a single slip set off a chain reaction — and the office only found out once the ground had already gone quiet. We connected the field to the plan, and started seeing the stalls coming.
Run dozens of infrastructure projects at once, with hundreds of subcontractors, and the domino effect stops being a metaphor. It becomes a daily risk — and this company was paying for it in chronic delays and heavy late-completion penalties from government clients.
The problem wasn't the ground. It was a data blind spot. One subcontractor slipping — earthworks or drainage running late at a single site — jammed the next crew in line, threw out the whole project gantt, and triggered penalties. Project managers at head office only discovered the gap when a fresh crew turned up to a site that wasn't ready for them, which meant expensive idle days, disputes, and demands for extra payment.
Underneath it all, the information was trapped. Site managers' daily reports went out over WhatsApp or on paper, and never synced back to the planning and control tools at head office. The people managing the schedule were working from a picture that was always a few days stale.
Daily field reality, matched against the plan — weeks ahead of the jam.
We connected the digital work logs, subcontractor attendance and schedules from all 35 sites into a single, continuously synced control board — so head office finally saw the same reality as the field, on the same day.
An engine compares the true rate of progress at each site, every day, against the planned gantt — turning a pile of daily reports into a live read on where the project actually stands.
The model forecasts a stall before it happens — for example, “the piping subcontractor on the Route 44 site is running 20% below pace, which will delay the start of paving in about two weeks.”
The moment a delay is predicted, the system recommends an updated gantt and a shift of crews or heavy equipment from a site that's free — so capacity moves to where it's needed instead of standing still.
Nothing reshuffles on its own. The system sends a short confirmation to the project manager's phone; a tap approves the change. The AI flags and recommends — the manager commits.
Instead of chasing status site by site, managers see the whole portfolio in one view — every crew, every dependency, every risk — and manage by exception rather than by rumour.
The delays they could see coming were the delays they could stop.
Schedule overruns and delay days on infrastructure handovers fell by 35% — and the late-completion penalties and idle-crew overheads that used to eat a project's margin were largely eliminated on every active site.
Project managers moved from managing crises and arguments in the field to running work that was predictable and planned in advance — the same shift, with the same weeks of head-room, on every project in the portfolio.
The prediction was only possible because the field and the office were finally on the same live picture — the unglamorous integration work is what made the forecasting real, the same way our workflow-automation work bridged a legacy system before any AI could touch it. And because a human approves every reschedule, the crews trust the plan they're handed. Foresight, not autopilot.
Reef TRH