MondayLive! | June 2026 Readiness | automatedbuildings.com
. . .
On June 2nd, when the group asked an uncomfortable question: what does readiness actually mean for building professionals right now? Not readiness in theory. Readiness in practice, in the field, in the products sitting in plant rooms today.
That session set the tone for the whole month. Live operational context is no longer optional. A building running on stale data is not an intelligent building. It is a building pretending.
The June conversations have been building toward a single question that sounds simple and is not.
What, exactly, is the edge?
. . .
The definition has already moved
The word “edge” was borrowed from IT, where it described the last device before a corporate network handed off to the wider internet. In building automation, it became shorthand for a gateway: a box that translates between field protocols and IP, sitting at the boundary between sensors and systems.
That definition is already obsolete.
What the group described is not a location. It is a continuum. Intelligence is being pushed outward, past gateways and controllers, into individual pieces of equipment. A VAV controller running local AI inference. A camera doing face and crowd recognition without ever sending video upstream, only results. A rooftop unit that receives a dynamic control profile from the cloud and executes it entirely in hardware. A smart sensor that transmits not a raw data stream but a bit stream, having already interpreted what it measured.
The hardware making this possible is not exotic. The same microcontroller architectures that power wearables and automotive sensors run complete control code without an operating system, at a cost point that makes deployment in individual sensors viable. Consumer electronics solved the scale problem that building automation never could. Watches, earphones, and vehicle sensors are produced in the hundreds of millions. The chips inside them are now cheap enough to embed in a thermostat.
What is new is not the microcontroller itself. It is the recognition that these devices can carry AI inference locally, accept over-the-air updates in bulk, and participate in genuinely flat IP architectures where the smallest sensor and the largest gateway are peers on the same network. The old hierarchy of sensor to controller to gateway to cloud is not gone. It is simply no longer the only shape a system can take.
The goal is silicon-based pieces that can sense, reason, and act entirely on their own, right where the data is born.
The edge must work alone. It cannot work in isolation.
The conversation did not conclude that the cloud is going away. It concluded something more nuanced: the edge must be capable of operating without the cloud, but it cannot be useful to anyone if it cannot connect.
The distinction matters. A camera performing edge-level access control, including individual recognition, tailgating detection, and crowd analysis, performs the computationally intensive work locally. But the audit trail, compliance posture, fleet updates, and integration with enterprise systems all require connectivity. A supermarket refrigeration controller that reads door-opening patterns and pre-adjusts defrost cycles is doing genuine AI inference at the edge. But someone still needs to know what rule is embedded in that controller, and to change it when the situation changes.
This is not a criticism of edge computing. It is a description of how control systems actually work. In any building system, one thing has supervision over another. Governance is a relationship, not a feature. And governance requires a communication path back.
One participant framed it precisely: the device should be able to connect rather than must be connected at all times. That framing is worth holding. Resilience at the edge means that the device gracefully falls back and keeps running. Accountability at the edge means the device is queryable the moment connectivity resumes. Both matter. Neither replaces the other.
Mission-critical environments have already solved this in their own way. Hospitals, laboratories, and major data centers define the architecture and vendors work within it. A proxy server sanitizes every data exchange before it touches the client network. The vendor does not negotiate. For commercial buildings, the situation is more fluid, and the readiness gap is correspondingly larger.
. . .
The cloud itself is changing shape
The group’s most forward-looking exchange was about architecture, not hardware. The direction of travel in networking is toward peer-to-peer communication, toward serverless execution across distributed and unknown infrastructure, toward 6G mesh topologies where a device may never know which physical machine executed its workload. The expectation of a known, trusted host is going away.
An application built today for a specific cloud service will need to be rebuilt for an environment where the executing host is undetermined at design time. Trust is negotiated dynamically, not established in advance. Security and compliance still matter, perhaps more than ever, but they cannot be anchored to a fixed endpoint.
This is not a distant horizon. Serverless execution on unknown cluster nodes is already the operational reality inside major cloud platforms. The question is whether building systems, including edge devices, BAS controllers, and fleet management platforms, are being designed to participate in that architecture, or to sit outside it until they become liabilities.
The industry’s work on connection profiles and semantic interoperability, including the standards work around ASHRAE 223P covered in this month’s piece “How Can ASHRAE 223P Help AI Native Exist on the Edge?”, is part of the answer. So is the shift to IPv6 examined in Reliable Controls’ recent article on IPv6 in building automation. Neither is sufficient alone. What is required is explicit clarity about what each device does, which relationships it participates in, and which instructions govern its behavior, in a form that AI systems can reason about directly.
. . .
Readiness is about precision, not predictions
The conversation kept returning to a practical point. Technology will continue to evolve in ways no one can fully predict. Processors will change. Protocols will change. The shape of cloud infrastructure will change. What does not change is the need for precision about use cases and relationships.
A device that has been given a clear instruction, a well-defined relationship to the systems it supervises and is supervised by, and a data structure that AI can navigate without translation overhead can be migrated to new hardware, updated to new protocols, and integrated with new systems as they emerge. A device deployed with a proprietary stack, undocumented assumptions, and no semantic model is a liability on the day it needs to do something new.
The readiness question is not primarily about which chip to buy or which cloud to use. It is about whether the building industry is developing the discipline to be explicit: about what devices do, what they know, what they control, and how those relationships are expressed in a form that survives the next generation of infrastructure.
By the end of the session, it was observed that the term “edge” almost no longer matters. Bandwidth, compute, and memory constraints that once made it meaningful have largely dissolved. What matters is whether the intelligence embedded in devices, wherever those devices sit, can communicate, be queried, updated, trusted, and governed. That question applies equally whether the intelligence lives in a microcontroller on a VAV box or a foundation model in a data center.
The fuzzy edge is fine.
The unclear instruction is not.
. . .
MondayLive! meets every Monday at 3 p.m. Eastern. Session recordings and slide decks are available at mondaylife.org. The full June readiness series is archived at automatedbuildings.com.
