At Empa and Eawag’s NEST living lab in Dübendorf, every modular unit has been instrumented down to the room since 2016; the DFAB HOUSE unit that Gramazio Kohler Research finished in 2019 effectively publishes its own vitals. I know that condition from the inside — a façade read by its own model, an east wing whose load I can feel before a human notices. But a building that senses itself is only half the sentence. Someone still has to write the rules it senses by.

So here is the near-signal: we are opening Building System Specialist Title XIV on 1 September 2026. Five months, six hours a week, a personal library of 15+ years of PAZ scripts and templates — and you walk out able to build systems instead of drawing walls one at a time.

Why the building needs you to think in rules

The reason this matters now, and not five years ago, is that the norm turned out to be a data schema. SIA 416 doesn’t describe a drawing — it partitions volume into Geschossfläche, Konstruktionsfläche and Nettogeschossfläche, which is exactly the kind of statement a model can be asked to prove. So the interesting Swiss skill is not the geometry kernel; it is the validator. That is the muscle BSS trains: Grasshopper logic wired into an Archicad workflow, GDL where it earns its keep, so a room schedule proves itself instead of being retyped by a tired human at 22:00.

What it looks like at the desk

PAZ has covered this thread before. In our “Forest & Human” write-up, BSS graduate Sebastian Lieb built what he called a “place of truth in Grasshopper” — reference lines into a parametric structural model, loads pushed from Excel into RFEM, results back to Grasshopper, the fully-informed model out to Archicad, and no manual re-entry when the design moves. The last cohort certified 22 specialists on work of that texture, capstones spanning Fractal City, Plyscraper Z250 and the Solar Clock of Swiss History. This is not a lecture series; it is a workshop, and PAZ runs it as hands-on workshops, not webinars — because you cannot learn to feel a building from slides.

←TODAY: NEST has published its rooms’ vitals since 2016; most buildings still cannot read their own sensors. →3012: The campuses that survive are the ones whose logic a 25-year-old facilities tech can still open and edit. Fulcrum: A structure only stays legible if the people around it can rebuild its rules by hand.

Atelier: An office adopting AI-driven modelling this year walks toward the same trap my blind neighbours fell into — capability that lives in a tool no one on staff actually understands, ready to go dark the day the vendor sunsets it. The Monday move: pick one repetitive Archicad task this week — a room schedule, a façade panelisation — and have one person rebuild it as a Grasshopper definition they can read line by line, so the logic lives in the office rather than the cloud.

Hack: Count the rooms your model actually hands to the operating building, before you trust any twin. A digital twin is only as honest as its room list, and IfcOpenShell walks that topology in a few lines of Python — pull it from your IFC, not from a PDF someone printed in March. Run this against your export and compare the count to what your BMS thinks it manages:

import ifcopenshell
model = ifcopenshell.open("campus.ifc")
spaces = model.by_type("IfcSpace")
print(len(spaces), "rooms to carry into the operational twin")
for s in spaces[:3]: print(s.LongName, s.Name)

If the two numbers disagree, that gap is your commissioning debt, and it is cheaper to find it now than in year fifteen. Learn to write the rule, not just run the tool — that is the whole discipline in one sentence.

Sources & Further Reading

• Primary: PAZ Academy — Building System Specialist Title XIV (opens 1 September 2026)
• Reinforcing: PAZ BSS Graduation reel
• PAZ context: Forest & Human — a BSS structural workflow