The last MondayLive! session before the summer hiatus skipped the usual single-topic format for something more reflective: a look back at the first half of 2026 and a new framework for thinking about where the industry goes next.
A Year in Themes
The group walked through six months of monthly themes, each one building on the last. January looked ahead to the AHR Expo in Las Vegas. February reflected on what actually happened there. March took on governance, a topic the group worried would be dry but found unexpectedly rich. April mapped AI onto the layers of the “smarter stack,” the framework the community built a few years ago to describe how intelligence moves through a building’s systems. May returned to fundamentals, examining what the basics of the past mean for the future. June, the month just ending, focused on AI readiness.
Taken together, the arc traces a deliberate path: from logistics to reflection, to governance, to architecture, to fundamentals, to readiness. The next stop is autonomy.
The Trust Matrix
The session’s centerpiece was a new framework for understanding how trust shifts as buildings transition from human-operated to agent-operated. It is organized as a two-by-two matrix.
One axis runs from policy at the top to operations at the bottom. Policy is the layer that decides what should happen, the rules of engagement across a building’s systems. Operations is the runtime layer, where systems actually exchange information and act.
The other axis runs from human answerable on the left to autonomous agent on the right, representing a rough timeline of where the industry has been and where it appears to be heading.
In the top left quadrant sits the world most buildings still operate in today: policy set by humans, executed by humans. Trust here rests on reputation and accountability, mechanisms the industry understands well, even if they do not scale easily.
Moving right, the framework anticipates a quadrant where policy is still written by humans but executed by agents. The bottom right, where both policy and execution shift to agents, was flagged as the highest-risk zone and the one with the most open questions.
Where Are the Policies Written Down?
The discussion that followed the framework was arguably more useful than the framework itself. The group kept circling one question: if an agent is supposed to follow policy, where does that policy actually live?
Large organizations often have well-documented policies for HR, finance, and marketing. Building automation policy is a different story. Several participants pointed out that most facilities never wrote down a formal policy for who can access systems, how alarms should be handled, or what a sequence of operation should actually enforce. ASHRAE Guideline 36 was raised as an example. It exists as a published standard for high-performance HVAC sequences, yet adoption across real buildings remains inconsistent, and an agent has no way of knowing whether a given building is even attempting to follow it.
That gap, between what should happen and what actually happens, was named directly during the conversation as the real work ahead. Agents do not get distracted, do not skip steps, and do not quietly ignore a rule they disagree with. But that only matters if there is a rule to follow in the first place, and right now, much of the industry is operating without one written down anywhere an agent could check.
One participant offered a comparison to the auto industry: a poor human baseline makes even modest automation look dramatically better. The same logic was applied to building operations. The standard to compare against is not a perfect ideal, it is the inconsistent, undocumented status quo that already exists.
AI as Policy Archaeologist
A more optimistic thread emerged near the end. Rather than treating AI purely as a risk to be governed, several participants suggested it could help solve the very problem it creates. Reviewing, summarizing, and reconciling scattered policy documents across vendors, contracts, and departments has always been valuable work. It was simply never cost effective to do at scale with people. AI changes that equation. The technology that raises new governance questions may also be the most practical tool for finally answering them.
It was also noted, almost in passing, that the term “policy” already has a precise meaning inside AI systems themselves, where a policy is the model that governs an agent’s decisions. The industry is not inventing a new concept by applying the word to buildings. It is borrowing one that has existed in machine learning for decades.
What’s Next
MondayLive! returns September 14, the Monday after Labor Day, picking up the second half of 2026 with autonomy, trust, and the policy gap as open threads to develop further ahead of the ASHRAE conference in Chicago.
Have a great summer!
. . .
This post is based on a discussion from MondayLive!, a weekly online session held on Mondays at 3 p.m. Eastern. Visit mondaylive.org for session details and automatedbuildings.com for ongoing coverage of the smart buildings industry.
Further Reading
- How Can AI Help Us Organize and Create Policies for Building Operations? on automatedbuildings.com
- C4SB Cyber Security for Buildings Working Group on c4sb.org
- C4SB BACnet Cloud Collective Working Group on c4sb.org
- C4SB Open Controls Working Group on c4sb.org
- ASHRAE Guideline 36, High-Performance Sequences of Operation for HVAC Systems
