Technical Article: AI in Major Projects: From Hype to Real-World Utility in Project Controls

Artificial Intelligence (AI) has arrived—and it's not just penning poems for my daughter anymore. From advisory boards to site offices, AI is edging into our workflows, whispering promises of insight, automation, and foresight. But for the project controls professional, grounded in precision and governed by tangible outcomes, one question keeps surfacing: Is AI practically useful now, or still trapped in a cycle of hype? My view? It’s here in narrow but powerful ways—and it’s changing how we think about effectiveness, risk, and delivery certainty. But we need to move past awe or anxiety and start using it with clarity and purpose.

What Is AI (and What Isn’t It)?

AI is not a singular monolithic technology—it’s a family of systems with different applications:

• Predictive AI:Learns from past data to make forecasts. This is already being used in schedule risk analysis and performance estimation.
• Generative AI: Creates new content—text, code, even images—based on patterns. Useful for reports, summaries, and even stakeholder comms.
• Agentic AI: Takes autonomous action toward defined goals. Early days here, but it could soon draft updates, flag risks, or initiate tasks without being prompted.

Most of the AI tools in project environments today sit in the predictive or embedded generative category.

Use Cases: Where AI Meets Project Controls

Before getting starry-eyed about toolkits, it’s worth anchoring on how AI can improve project control outcomes. Some high-impact uses include:

• Schedule Risk Forecasting – Identifying slippage drivers early
• Estimate Optimization – Learning from past project curves and benchmarks
• Program Sequencing – Generating build logic options at scale
• Reporting Automation – Drafting dashboards, briefs, and status summaries
• Anomaly Detection – Spotting outliers in spend, progress, or performance
• Decision Support – Synthesizing multiple data points for complex choices

Each of these aims at a familiar outcome: less rework, better foresight, and fewer surprises.

Types of AI Tools in Projects

Let’s group the available tools to make sense of what’s really out there:

1. Purpose-Built, Embedded AI Platforms

These integrate predictive or generative AI into project delivery tools.

• Nplan: Learns from thousands of schedules to assess delay risks.
• Nodes & Links: Evaluates schedule health and dependencies.
• ALICE Technologies: Builds and tests thousands of sequencing options instantly.
• Octant AI: An Australian-born platform, using predictive analytics to improve capital productivity by measuring and reducing overrun potential in real time.

2. General-Purpose AI Assistants

These aren't built for projects but can be adapted with prompting frameworks.

• Chat GPT, Copilot, Perplexity, Rovo: Can draft communications, simulate stakeholder updates, and even create scenario analyses if guided well.

3. GBuild-Your-Own AI

For organizations investing in tech stacks, this includes:

• Calling LLMs (Large Language Models) via APIs
• Building AI agents that integrate with project dashboards or controls workflows
• Customizing enterprise AI copilots for cost, schedule, or document control

Crafting Better Prompts: The CRAFT Framework

To unlock real value from generative AI—especially the general-purpose kind—we need better prompts. Here’s a simple structure I find helpful:

CRAFT:

• Context – What is the project background?
• Role – Who should the AI “be”? (e.g. Scheduler, Risk Advisor)
• Action – What exactly should it do?
• Format – Report? Email? Table?
• Tone – Formal? Plain-English? Advisory?

Example prompt:
"Act as a senior risk analyst reviewing a $2B transport megaproject. Identify the top three drivers of schedule overrun based on these indicators. Present in a one-page executive summary.” This improves output quality and helps establish trust by reducing ambiguity.

The Disruption is Real—But Not Universal

Let’s be honest: “The bots aren’t taking your job” isn’t the whole story. Repetitive tasks, data analysis, even communications—these are already being transformed. Job functions will shift. Roles will be redefined. And yes, for some, that will mean disruption. But for most professionals, the opportunity lies in augmentation—freeing up time from grunt work to focus on strategic judgment, stakeholder alignment, and value engineering.

Where to From Here?

We’re still early in the curve. The real disruption will come from:

• Systemic integration of AI into governance frameworks
• Shift from tools to agents — where AI doesn’t just answer but acts
• Greater emphasis on prompt literacy and data stewardship

We need more professionals who can curate and interpret AI output, not just consume it. That’s where leadership lies.

Final Word: Lead with Curiosity, Act with Precision

I use AI every day—not because it’s trendy, but because it’s useful. Whether simulating risk, summarizing board papers, or translating technical insights into plain English, it helps me deliver better. We don’t need to blindly follow hype. But we also shouldn’t wait until the tools are perfect. Let’s lead—critically, intentionally, and transparently. Now is the time to shape how AI is adopted in project controls—not after it’s shaped us.