Meet your new building team: People, machines, AI agents, buildings

Our recent webinar brought together four industry pioneers to demonstrate how open standards, semantic data models and AI are breaking down information silos to create truly intelligent buildings that can communicate, optimize and verify their own performance.

The panel:

Stephen Lothrop, Lead Electrical Engineer at PAE

Kimon Onuma, FAIA, President of ONUMA, Inc. and Founder of BIMStorm (Moderator)

Tristan de Frondeville, Founder & CEO of SkyCentrics

Zahra Ghorbani, PhD, Senior Digital Design Manager at Massport

Watch the full Webinar

The challenge

Buildings today are drowning in disconnected data. Even award-winning "smart buildings" like PAE's Living Building in Portland have their intelligence trapped in silos: BIM models, building management systems, Excel schedules and PDF wiring diagrams that can't talk to each other. As Stephen explained, "Even though we knew where everything was, it was scattered all over the place."

The result? AI tools confidently provide wrong answers because they lack building context. Ask about wastewater handling, and AI assumes standard municipal sewers, missing entirely that PAE's composting toilets create sellable plant food in the lobby.

The solution: Knowledge graphs + open standards

The team demonstrated a breakthrough approach using semantic data models that create a "knowledge graph": an interconnected web of building information that AI can actually understand. By leveraging open standards like ASHRAE 223P, BRICK ontology, Project Haystack and Real Estate Core, they've created a common language for buildings.

Kimon drew a powerful analogy: "My phone knows my heartbeat, location and has a unique ID anyone can call worldwide. We're not reinventing technology; we're applying what already works to buildings."

The game-changer? These systems now enable natural conversations with buildings. Tristan demonstrated asking, "This room is hot," and the AI instantly analyzed the knowledge graph to identify which systems control that space and suggest solutions.

Key breakthroughs

The maturity framework: Zahra introduced a comprehensive assessment tool developed with NIST that scores digital twin systems across 22 dimensions in four domains. "We can't improve what we can't measure," she noted.

LEED v5 automation: The team demonstrated automatically evaluating PAE's building against LEED v5 requirements by querying the knowledge graph with no manual checklists required.

Linux-powered resilience: SkyCentrics' approach uses off-the-shelf Linux computers with N+1 redundancy, something unheard of in traditional building management systems.

Cybersecurity by design: NIST is actively testing cybersecurity protocols on this very building, with modern containerized applications providing security layers that legacy BMS systems never contemplated.

Key takeaways

The phone number principle: Just as international calling standards enable global communication, buildings need standardized IDs and data structures to enable interoperability.

Start with what you have: PAE took existing Revit models, Excel schedules and system data, then transformed them into 3,000 identified assets with nearly 200,000 semantic relationships.

AI needs context: Without semantic models, AI makes confident but wrong assumptions. With knowledge graphs, it provides specific, accurate answers grounded in actual building data.

Open beats proprietary: The Coalition for Smarter Buildings and Linux Foundation approach ensures no vendor lock-in, enabling rapid innovation across the industry.

The tools are here. The pathway from siloed building data to conversational AI is now more accessible than ever.

Want to see these technologies in action and connect with industry leaders? Join us at Greenbuild 2026 in New York, including the full-day AI Summit.