Marjan Gholamalipour is Americas West Program and project management business leader at London-based design and engineering consulting firm Arup. Markus Bernhardt is principal at Walnut Creek, California-based artificial intelligence strategy consultancy Endeavor Intelligence. Opinions are the authors’ own.
The conversation around artificial intelligence in project management is clamorous. A deluge of new tools promises to help individuals draft emails, summarize meetings and analyze data faster. While useful, this focus on personal productivity often obscures a more profound shift. To find the real value, we must look beyond the immediate tools and toward the emerging systems that will change how we govern project delivery.
Deepening AI integration
For most project professionals today, AI shows up as personal augmentation. This is the immediate impact of tools such as Microsoft Copilot, which allows us to use generative AI to draft a project charter or analytical AI to create a quick cost forecast. The outcome is personal efficiency. These tools make individuals faster, but they do not fundamentally change a team’s underlying processes.
Widespread adoption of these tools is essential to build literacy, but it is merely the foundation for the next, more powerful application of AI: a shift from personal efficiency to systemic intelligence. Systemic intelligence embeds AI into the governance of a project, program or portfolio. Instead of simply accelerating individual tasks, the system delivers intelligent signals that enable decision makers to act promptly and with confidence. The key distinction is this: does the AI merely provide data for later analysis or does it help project managers with decision making in real time?
Focusing on what matters most
Implementing systemic intelligence requires us to move beyond the traditional dashboard. While dashboards visualize data for human interpretation (“look what happened”), systemic intelligence actively uses data to prioritize attention (“decide on this now”).
Consider a complex capital project where the supply chain is integrated into a systemic model. An AI agent, analyzing logistics data and weather, might detect a probable 72-hour delay for a critical steel delivery. In the traditional model, this might surface later during a status update. In a systemic intelligence model, the system compares this signal against an agreed tolerance. Because the delay exceeds that threshold, it automatically opens a time-bound decision window for the procurement manager. The choice presented is clear and accountable: approve expedited shipping for a 15% cost increase or formally accept the schedule slip.
In such a case, the system did not just provide data, it forced a timely, auditable decision. This intelligence is typically triggered by signals falling into specific categories, such as cost forecasts, clash detection anomalies or contractual obligations. Each signal is a machine-generated insight that, by crossing a pre-set line, demands human judgment.
For experienced project managers, this approach codifies the intuition they have used for decades. We often apply implicit tolerances to our projects, knowing intuitively which delays matter and which are noise. Systemic intelligence makes these tolerances explicit. This systematic filtering frees human attention from chasing minor past variances and focuses effectively on the moments when a forecast error signals a genuine need for intervention.
The operational impact: moving from reporting to deciding
When systemic intelligence is in place, the rhythm of management changes. The weekly status ritual, in which teams spend hours narrating what happened last week, transforms into a forum for decision-making. Time traditionally spent on narration is reclaimed, and sessions become focused on clearing the specific decision windows that the system has opened.
Ideally, the system analyzes the cost, schedule and resource impacts of a delay, presenting the project manager with the top three mitigation options. The human role is elevated from problem-finding to using tactical judgment to make the final call. This clarity reduces the need for escalation and enables a true “eyes on, hands off” leadership approach.
Common pitfalls: why AI initiatives stall
When do these initiatives fail? A common pattern is “pilot theater.” This occurs when a project team runs an impressive AI pilot that generates signals, but those signals are not connected to any formal decision point, so nothing actually changes, and enthusiasm drains away. Another barrier is the black box problem. When stakes are high, a leader will rightly refuse to act on a warning if the system’s reasoning is opaque. Trust is the currency of project delivery. If the AI cannot explain its provenance, specifically which data sources it used and why, it will be ignored.
None of these are purely technology failures. They are gaps in governance, process design or data quality.
Foundation first
Ultimately, the transition to systemic intelligence is not just a software installation. It is a redesign of how we value human judgment. The journey begins not with a massive technology investment, but with a simple question: “What is the most important decision we have to make this week, and what single piece of data would help us make it better?”
However, this intelligence cannot exist in a vacuum. It requires a foundation of truth. If our project data is trapped in disconnected files and digital silos, no amount of AI can save the schedule
