What can adventures in vibecoding and broadening one’s understanding of AI models reveal about the future of infrastructure? In this episode, the podcast makes a 180 as Lawrence Rowland turns the microphone on Riccardo. What follows is a wide-ranging and deep-delving conversation about modelling, artificial intelligence, systems thinking, and the role these evolving tools might play in major programmes. 

They begin with Riccardo’s development of an International Roughness Index (IRI) prediction model to better understand pavement degradation—an example of using data to challenge linear assumptions and improve infrastructure decision-making. From there, the discussion follows Riccardo’s learning journey through reinforcement learning, coding agents, decision intelligence, reality capture, knowledge graphs, and the future of physical AI. Along the way, Lawrence and Riccardo bring the discussion back to what these tools mean for project leaders: why humans and organizational processes remain the weakest link in adopting new technology, how AI can support—not replace—human decision-making, and why meaningfully connecting the data is the biggest determinant of success.

The episode also explores the softer but no-less-essential side of major programmes, including relationship complexity, stakeholder power, information asymmetry, and the limits of tools like Gantt charts in representing the real dynamics of project delivery. Finally, Riccardo reflects on what his AI experiments and podcast journey have taught him about his own niche. This conversation covers all the bases: curiosity, humility, and the need to keep rethinking how we understand, manage, and learn from complex infrastructure systems.

Key takeaways

• The future of LLMs and physical AI in major projects;
• What old-school video games have to do with building prediction models;
• How decision intelligence can begin to support, but not replace, project leadership;
• The difference between complex and complicated—and why it matters for systems thinking;
• What delving into AI and its possibilities has taught Riccardo about his niche.

Quote

•  “[In today’s LLMs], through brute prompting, you can actually get to the answer you want…but it’s a very convoluted path.” - Riccardo Cosentino

The conversation doesn’t stop here—connect and converse with our community via LinkedIn:

• Riccardo’s spreadsheet “AI Startups in Engineering & Construction”: https://docs.google.com/spreadsheets/d/1TCvfRqqd1NzxgMhVoTMbKeL9bOp_1-mn/edit?usp=sharing&ouid=100654060211988607651&rtpof=true&sd=true
• Riccardo’s LinkedIn Article “AI Startups in AE&C (Architectural, Engineering and Construction)”: https://www.linkedin.com/pulse/ai-startups-aec-architectural-engineering-riccardo-cosentino-bvhrc/
• Follow Navigating Major Programmes: https://www.linkedin.com/company/navigating-major-programmes/
• Read Riccardo’s latest at www.riccardocosentino.com
• Follow Riccardo Cosentino: https://www.linkedin.com/in/cosentinoriccardo/
• Follow Lawrence Rowland: https://www.linkedin.com/in/lawrencerowland/