The challenge
A regional developer running 12 simultaneous high-rise projects had no single source of truth for schedule, budget, and resources. Updates lived in spreadsheets. Critical-path replanning took 4–6 weeks. By the time the new plan landed, reality had moved on.
Every delay cost $120K–$400K per day across the portfolio. Every late MEP delivery rippled into structural and façade. Cost overruns showed up in monthly P&L — too late to course-correct.
Approach
We unified all projects on a shared event model: every task, contract, delivery, and inspection emits events. The AI layer sits on top, watching for patterns that human PMs miss: subcontractor delays correlating with weather, supplier lead-time creep, hidden dependencies between trades.
- Real-time Gantt across all projects · drag-and-drop replan that propagates dependencies in milliseconds
- AI agent suggests reallocations with a one-click "Apply" — every suggestion has an auditable rationale
- Cost-impact preview before any change · finance sees the P&L delta before approving
- Mobile-first for site teams · check-in, photo-evidence, daily reports run on phone
- Stakeholder dashboards — investors see what investors care about, contractors see their slice
Stack
Results
Twelve months in, the developer cut average schedule overrun from 23% to 5%. Cost overruns dropped 71% vs baseline. Resource utilization climbed to 94% as the AI surfaced idle crews across projects.
- Replan cycle: 6 weeks → 12 days · most replans now happen daily, automatically
- Daily P&L visibility: finance sees impact within hours, not at month-end
- Subcontractor disputes: down 60% — the audit trail makes most arguments moot
- One project shipped 3 weeks ahead for the first time in the developer's history