从互联系统到可组合建筑：行业讨论语义互操作与集成人才之变

Monday Live! | October 12, 2026 | Evolving Skill Sets: The Future of Integration

. . .

Integration has always been the craft of making one system talk to another, one project at a time. This week’s discussion asked what happens when that work becomes plug and play and who is needed once it does.

Shape is not enough. Meaning has to snap together too.

The group returned to the Lego brick image from last year. Bricks connect because their geometry matches, but a three-quarter-inch pipe might carry water or hold electrical conduit. Matching size does not tell you what a thing is or what it connects to. That meaning is the semantic model, and it is where the building design world and the operations world still talk past each other. Design tools capture how a building goes together. Far less of what each system means makes it through to operations.

That gap is also where industries are converging. The design side is moving toward knowledge graphs and lighter, API-based data exchange, and its language increasingly mirrors the work in C4SB. The takeaway: the open, semantic approach is not a niche idea, and others are arriving at the same place independently.

Standards are the seams where modules separate.

Common sequences of operation, guidelines for how systems are structured, and ASHRAE Standard 223P for how data is described all give natural boundaries for modularization. AI can help synthesize them quickly into the next generation of the workforce. The drag, as always, is legacy. Roughly 5.5 million commercial buildings were never set up this way, so forward-looking owners will need to prioritize their portfolios and modernize the data underneath them.

More choice is a different kind of complexity.

Plug and play makes assembly easier, but it multiplies options: which edge device, which data layer, and which application. The group agreed this is an opportunity, not a burden. Performance requirements and use cases will narrow the field, and weaker products will fall away. As with real Lego, you can build anything, including houses.

Do not let AI become the new lock-in.

If every vendor bundles its own proprietary AI and chips, the industry risks recreating the silos it just escaped. The answer is to decouple AI from the data. Data should be reachable by a person, a large model, or an on-site model, without passing through one vendor’s gate. The car dashboard analogy fits well: you do not need to understand every component, only that the common plug works.

The skills question: integration moves up the stack.

If moving data from point A to point B stops being the hard part, the work shifts to creating value from integrated data. That points to engineers who understand more than one domain, such as HVAC, lighting, and physical security, with AI helping fill knowledge gaps on demand. The group was candid about limits:

One person cannot master every domain, so the real change is team management. Technicians, analysts, and integrators need to share what each knows.
Budgets and KPIs are still siloed. Without top-down changes to how teams are measured, integration will not deliver its value.
Owners often do not know what to ask for. The industry has not shown them what is possible beyond one vendor.
Manufacturers have to build products that support these standards first. Training follows from there.
Technicians want training and will stay where they get it. Pairing experienced people with newer talent works better than sending newcomers to shadow one veteran.

Fifty years of technology transitions offered a calming perspective: core knowledge stays useful, and the people who know what not to do with a chiller or fan are valuable partners to those who know the new tools. What has changed is speed. The adaptation cycle is shorter, so continuous learning and change management matter more than ever.

Around the industry

Updates this week covered a standards summit where semantic models and knowledge graphs drew serious attention and an industry conference where upstream work won out over flashy dashboards. Validating data before deploying analytics and setting clear limits on what AI may control were repeated themes. C4SB also flagged data first, thinking of it as a coming focus: start with the data, not the application.

The closing challenge

The industry owes its owners better examples of what is possible so they can ask for the right thing. When they do, specifiers, manufacturers, and integrators can deliver it.

Further reading on AutomatedBuildings.com

The post From Connected Systems to Composable Buildings: Who Makes It Work? appeared first on AutomatedBuildings.com.

「英中对照」从互联系统到可组合建筑：行业讨论语义互操作与集成人才之变