This week’s MondayLive! Discussion delved into a fundamental challenge in smart buildings: the “superglued” digital twin. Many buildings today are collections of proprietary, siloed systems that function in isolation. While they may operate independently, their data and controls are locked away, preventing holistic optimization, true interoperability, and the effective use of modern tools such as AI.
The conversation presented a powerful alternative path: strategic decomposition and recomposition.
Drawing an analogy to LEGO bricks, the session outlined a methodical approach:
- Decompose the Digital Model: The process doesn’t involve physically deconstructing the building. Instead, it focuses on digitally decomposing the information architecture—untangling the data, points, and system relationships trapped within proprietary walled gardens.
- Identify & Map “Studs”: The core is identifying the essential connections (the “studs”), such as how a battery system relates to solar PV production and building load. This often reveals mismatched naming conventions and missing digital links between physically connected assets.
- Recompose with Open Standards: Using open APIs, semantic models (like RDF graphs), and cloud-native principles, these newly liberated “building blocks” are recomposed into a flexible, interoperable data fabric. This creates a single source of truth, with relationships between systems clearly defined and accessible.
The Tangible Outcome: A “Living” Building
This approach transforms a static collection of systems into a dynamic, “living” digital asset. The recomposed platform enables:
- Unified Visibility: Dashboards that combine data from previously siloed systems (e.g., energy storage, generation, and consumption) in one view.
- AI Readiness: Provides structured, contextual data that allows AI agents and natural language interfaces to deliver accurate, actionable insights instead of “hallucinating” due to poor data context.
- Future-Proofing: New applications and use cases can be built on top of the open data layer, protecting the owner’s investment and avoiding new vendor lock-in.
A Practical Path Forward
A key insight is that this isn’t an all-or-nothing endeavor. The strategy advocates starting with a minimum viable approach—decomposing and connecting the most critical assets and use cases first to demonstrate value. This makes the process achievable for both new construction and existing buildings, with the goal of bringing all systems to a common, interoperable middle ground over time.
Natural-language building dialogue is now within reach. The entire multi-year push toward decomposition, open APIs, and semantic graphs has reached the tipping point: reliable, hallucination-free conversations with buildings (“What’s using power on floor three right now?”) are no longer a distant dream but an imminent reality, driven by edge AI chips and connected graphs. This single capability will fundamentally change how every owner, occupant, and technician interacts with buildings. A modern flagship building had to be digitally torn apart and rebuilt in months. Even a five-year-old, award-winning net-zero headquarters from one of the world’s top engineering firms was found to be a collection of proprietary silos with inconsistent naming and zero usable relationships. The fact that this extreme (but successful) decomposition-and-recomposition effort was required on a “best-in-class” building proves the problem is universal and far worse than most admit—and that the fix works. Owners must demand digital deliverables at procurement or forever stay locked in. The clearest, most actionable lever emerged: every piece of equipment (new or retrofit) must contractually arrive with consistent naming, room/location mapping, and open APIs. Without this requirement written into specifications and contracts today, every project—regardless of how “smart” it claims to be—will immediately create new silos that will cost millions to untangle later. This is the one decision that decides whether a building ages into an asset or a liability.
This post used AI
watch the session here
