Project controls
Project controls in the age of AI
The PMO's job has always been to turn schedule and cost data into decisions. AI doesn't replace that judgement — it removes the compiling that gets in its way.
Project controls is one of the most data-rich disciplines in any organisation — thousands of activities, baselines, actuals, resources and risks, updated every period across a portfolio. And yet most PMOs still spend the bulk of their week compiling that data into reports, leaving too little time for the judgement the data exists to inform. That's exactly the imbalance AI and automation should fix.
Automate the compiling, not the judgement
The valuable work in project controls is human: interpreting why a schedule slipped, deciding where to intervene, having the hard conversation with a delivery lead. The low-value work is mechanical: pulling XER files, running the same DCMA checks, assembling the same board pack. AI's role is to take the mechanical layer off the PMO's plate so the judgement gets more of the week, not less.
Schedule quality, continuously
A DCMA 14-point pass by hand is a day's work, so it happens rarely. Automated, it runs on every update across every project — turning schedule integrity from a periodic audit into a live signal you can steer by. The same is true for earned value: CPI, SPI and BEI computed automatically and surfaced in executive-ready views, refreshed from every P6 and XER feed. That's the premise behind Primer.
The goal isn't a PMO that compiles faster. It's a PMO that stops compiling and starts steering.
From lagging reports to early warning
Traditional controls reporting is backward-looking — here's what happened last period. With live data and pattern detection, you can move earlier: flag the float eroding on a critical path, the resource squeeze building three months out, the baseline execution index quietly drifting below one. Early warning is worth more than precise hindsight, because it's still actionable.
Where to be careful
Project controls is high-stakes, so the same discipline we apply to any agentic AI applies here: guardrails, auditability and human sign-off on decisions that matter. AI should surface the exception and assemble the context; a controls professional still owns the call. Used that way — decisions augmented, compiling removed — AI gives project controls back its most valuable resource: time to think.