Twenty-Six Years of Chasing Interoperability

Editoral- by Ken Sinclair

Open AutomatedBuildings.com today and you will find an industry captivated by “agentic buildings”—AI systems that can reason across building data and act autonomously. Last month, Yves Habchy framed the remaining obstacle: vendor platforms remain fluent only in their native stacks, ignoring anything not tagged and modeled their way. The solution, he argued, is durable semantic tagging paired with independent integrators. Get that right, and proprietary platforms can come and go while a well-described building stays smart.

It is a compelling pitch, but also one I have published in various forms for decades. Having run AutomatedBuildings.com since 1999—and worked with pneumatic building control systems at Johnson Controls long before microprocessors arrived—I have watched multiple waves of “connect everything” promises crash and recede.

The Core Thesis:

Tracing the single thread of open systems, interoperability, and the fight against vendor lock-in across more than a quarter-century reveals a clear pattern: This industry has been directionally right about interoperability for twenty-six years, but almost always wrong about how close it was. Interoperability is not a destination; it is a moving frontier that simply climbs a layer higher with each generation.

Two Futures, One Issue (2000–2003)

The pattern was visible from the very beginning. In January 2000, two articles in the same issue offered starkly different paths forward.

One, written from the LonWorks camp at Echelon, declared the contest over: LonWorks had won, BACnet devices had proven non-interoperable in practice, and the industry should fall in line. The other, from Trane, argued “Why We Need Both,” framing BACnet and LonTalk as complementary standards destined to co-exist.

While BACnet ultimately became the lingua franca of building automation and LonWorks faded, the most accurate forecast was actually the most modest one. Jack McGowan refused to crown a winner, describing interoperability as a ladder. At the top rung sat true “plug-and-play” device substitution. McGowan predicted this rung would likely never be reached because manufacturers implement control logic differently. More than two decades later, he remains correct.

As the industry realized plug-and-play hardware substitution was a distant goal, the focus shifted to data unification middleware. By December 2003, Matrikon pitched OPC as the definitive savior to free integrators from vendor lock-in. Six years later, the same company was pitching OPC UA—a service-oriented architecture designed to bridge BACnet, LonWorks, and legacy systems to the enterprise. It was the same unifying promise, simply moved one layer up the stack.

Standardized vs. Interoperable (2008–2011)

As protocols matured, the industry frequently mistook a shared standard for true system-to-system compatibility.

In December 2008, BACnet International president Andy McMillan wrote about the alignment of BACnet and ZigBee, predicting standard wireless products by 2009. But he built in a crucial warning: labeling a product “ZigBee” did not make it interoperable, because the protocol only moved data without defining what that data meant. The real barriers were physical commission skills, battery life, and a reliance on trial-and-error debugging over genuine diagnostics.

This distinction—that a protocol is not an interoperability guarantee—recurs constantly. It surfaced in 2008 as a warning that “open” standards still permit proprietary objects, in 2011 when writers questioned if open protocols were ever enough, and in December 2025, when an industry roundtable targeted “the illusion of interoperability.” Contractors can easily check the BACnet specification box without ever enabling meaningful data exchange.

A Journey with No Arrival (2020–2026)

Whenever the industry nears interoperability at one layer, it redefines the concept to encompass the next hurdle:

• 2000: Controller-to-controller connectivity.
• 2009: IP convergence, with Contemporary Controls establishing IP as the backbone of choice.
• 2020: Security and encryption, solved on paper when BACnet Secure Connect brought authentication to the standard.
• 2026: Semantics and AI readiness—ensuring devices not only talk, but understand the context of the data they share.

Even our breakthroughs reintroduce legacy friction. Recent coverage of BACnet Secure Connect integration praises its security advances but cites immature vendor tooling, complex certificate onboarding, and the difficulty of bridging legacy hardware as persistent roadblocks.

The Cycle of Interoperability

By analyzing twenty-six years of industry discourse, we can map the predictable lifecycle of every “savior” technology:

What the History Tells Us

Applying this historical track record to today’s agentic buildings yields a clear, pragmatic forecast. The direction is correct: buildings will become more legible, automated, and data-driven, just as they eventually became IP-native. However, the fight against lock-in is simply migrating. Vendor AI platforms are poised to become the new walled gardens that independent integrators must pry open. Whatever timeline you are being quoted for the fully AI-native building, history suggests you should double it.

This perspective is not an invitation to cynicism. The pioneers writing in these pages over the last quarter-century have consistently pointed in the right direction; they have simply been endearingly impatient.

The ultimate value of this running archive is not just a record of past technologies, but a compass for future ones. It reminds us that real progress is a continuous road, not a final destination. As we begin writing the next chapter—agentic buildings and the semantic layer—this shared history remains our most reliable guide for the path ahead.

Sources

In order of appearance. All from AutomatedBuildings.com.

1. Tennefoss/Echelon, “Implementing Open, Interoperable Building Control Systems” (Jan 2000)
2. Hess/Trane, “Why We Need Both” (Jan 2000)
3. McGowan, “DDC Networks: Is the Question Really BACnet or LonWorks?” (Sep 2000)
4. Kondor/Matrikon, “Integrating OPC into Building Automation” (Dec 2003)
5. McMillan, “A Light at the End of the Wireless Tunnel” (Dec 2008)
6. Thomas/Contemporary Controls, “Connecting BACnet Devices to an IP Infrastructure” (Mar 2009)
7. Leonard/MatrikonOPC interview (Jun 2009)
8. Sinopoli, “How Buildings Will Communicate With the Smart Grid” (Dec 2009)
9. McMillan interview, “BACnet/SC Interoperability Acceleration Program” (Feb 2020)
10. McMillan, “Interoperability Writ Large” (Apr 2022)
11. “A Blueprint for HVAC’s Next Era (No Dumb Buildings)” (Dec 2025)
12. “Learn How BACnet Secure Connect Is Changing System Integration” (Dec 2025)
13. Habchy, “The BACnet/MSI Semantic Gap” (Jun 2026)

Brian Collins, Contributing Editor, inspired this post. Using today’s tools to unravel our past and understand our future. Thanks for the nudge. We would love to have you back, actively posting.

As self-appointed historian, I, too, seek to understand. I find myself these days much like you, reading AutomatedBuildings.com, to better understand our journey. Keeping our discussions open-ended and available to everyone is how we will achieve our future.