When “Too Complex” Isn’t True: Turning Live Building Systems into an Operational Digital Twin in Days
You’ve been told the complexity is non-negotiable. Before you can achieve true operations, you must first endure a months-long, monolithic project to build a perfect model. We took a different path. We started with live data, a single open API (Application Programming Interface), for software systems to exchange data directly from Facil.ai, and one simple question: Could we protect the ice cream freezers? We didn’t rebuild the system. We didn’t wait for a coordinated model. We layered a fully operational digital twin in days. The tools exist. The technology is here. What are you waiting for?
From anomaly to work order, in one connected flow.
1. The Myth Has Been Repeating
For years, we’ve heard the same assumption: building an operational digital twin takes a long time, especially for existing buildings. You have to reconcile drawings, clean point names, inventory assets, align vendors, and build integration layers before anything meaningful can operate. It’s framed as unavoidable complexity, the price of getting serious.
But what if much of that delay is habit rather than necessity?
Operating Reality: Facil.ai data viewed in Onuma
2. Start With What You Actually Have
A week before this year’s AHR, the largest HVAC and building systems gathering in North America, in Las Vegas, we tested that question in a live operating building. There was no coordinated BIM (Building Information Model). No polished handover package. No carefully structured lifecycle dataset waiting to be activated. What we had was live sensor data, an open API, and a willingness to connect.
Seven days later, we had a functioning operational digital twin: geometry grounded in context, live endpoints attached to identifiable assets, and rules capable of triggering maintenance logic.
And this is not just about existing buildings. New buildings become existing buildings the moment they are occupied. If we cannot bring them into operational state quickly, we simply defer the same problem.
At AHR we showed the deep reconstruction path at the PAE Living Building. This time we started from almost nothing but live data. Two extremes. Same direction.
What we actually had: sensors, an API, and momentum.
3. One Open API Was Enough
Facil.ai was already doing meaningful work. They connect to live systems, BACnet, Modbus, and proprietary protocols, and use AI to learn behavior quickly, then dynamically balance loads rather than relying on static schedules.
In some cases, they report energy savings of up to 48%.
They weren’t waiting for a digital twin; they were already operating intelligently.
Keith Gipson permitted us to use live data for the demonstration. On our first call, he exposed roughly 130 endpoints through an open API.
That was enough.
Thanks to Ari Gipson, who helped save the ice cream. The “protect the ice cream” use case didn’t come from theory. It came from the Gipson family’s real-world experience operating an independent ice cream shop and factory, where proper temperature control meant protecting product and profit.
When we needed fast reconnaissance of the building, Ari captured the Matterport scans and floor plan images that helped us build a minimum viable BIM in hours. From ice cream shop to digital twin, same mission. Protect the product. Make it work.
4. The Ice Cream Use Case
We chose a simple use case: ice cream freezers. Most people assume the danger is melting. Keith corrected us, freezer burn from units running too cold, especially during defrost cycles, is often the real issue.
We asked a straightforward question: could we locate those freezers, connect their live temperature data, visualize behavior, and define actionable rules?
We didn’t start with perfect labels. Engineers recognize patterns. Temperature ranges tell their own story.
Phone scan. Photo. Map pin. Spatial grounding in hours.
5. No BIM? Build the Minimum
There was no coordinated model. So we created a minimum viable BIM. A site visit by Ari Gipson using a mobile phone and a Matterport scan. A photographed plan. A quick geospatial check. Within hours, we had spatial context.
The Facil.ai Gateway was discovered midstream and added instantly.
6. Adjust Live, Don’t Rebuild
We linked endpoints directly through the API.
When we needed time-series access, the API evolved.
When we discovered the gateway location, we inserted it live.
Nothing stopped. Nothing was rebuilt. The digital layer kept pace with operations.
Each system does what it does best: sensing, deciding, acting.
7. From Sensor to Action
With temperature ranges defined by Keith, real numbers, not abstractions, we embedded alert logic. If a freezer drifted out of range, a work order could be triggered.
One freezer. One rule. The pattern scales. In just over a week, including coordination and iteration, we built a working operational digital twin. With prepared APIs and repetition, the first version could be assembled in hours.
The bottleneck is not technology. It’s a mindset.
8. This Isn’t About Retail. It’s About Pattern
This happened at Northgate Market, and they deserve the credit. But it could have been Northgate Airport, Northgate Hospital, or South Any Building. The name doesn’t matter. The pattern does. Start with live data, add context lightly, avoid monolithic rebuilds, and move in days instead of months. Every building ends up in operations. The approach works for all of them.
Deep reconstruction or rapid layering both lead to operations.
9. The Bottleneck Isn’t Technology
We didn’t replace Facil. We didn’t rewrite their algorithms. We didn’t demand upfront adoption of a specific standard. Facil’s AI continues to do what it does best, learning and balancing loads. As Keith put it, the AI is relentless. We simply layered spatial grounding and structured relationships on top.
Which means this isn’t a technology problem.
The tools exist. The APIs exist. The pattern works.
This is no longer a technology constraint. It’s a governance decision.
When the barrier is no longer technical, leadership becomes the lever.
A week of work. A live conversation between Keith Gipson and Kimon Onuma.
10. Retire the Myth
There’s noise about SaaS fatigue and integration overload. The issue isn’t SaaS. The issue is closed systems that block data and make simple connections hard.
Developers want clean APIs and stable identifiers. That’s what made this possible.
We demonstrated deep semantic reconstruction at PAE.
We demonstrated rapid connective layering here.
Two extremes. Same direction.
The technology is here. The myth that it takes months to begin should be retired.
This article didn’t unpack every technical layer or standards detail. That work exists elsewhere. At the end of the day, if the mission is to protect the ice cream and it works, the technology should disappear into the background.
That’s the point.
A week of work. A live conversation between the building and AI.
AHR Series: Four Conversations, One Direction
At AHR in Las Vegas, the conversation moved from theory to execution. These four pieces trace that shift, from platform tension to real-world proof to a coalition building the bridges at scale.
Part 1-When Platforms Stop Fighting and Start Connecting
From platform wars to platform peace, why interoperability requires alignment, not dominance.
Read here.
Part 2-From Models to Meaning: Bridging the PAE Living Building
The PAE Living Building as proof that BIM, semantics, and operations can stay connected.
Read here.
Part 3-A Coalition in Motion: Building the Bridges at Machine Speed
Standards, code, and leadership converging at machine speed.
Read here.
Part 4 – The Proof is in the Ice Cream
From live sensor data to operational action in days, proving complexity is a choice, not a constraint.
Read here.
