On 1 September 2026, PAZ Academy opens Edition XIV of the Building System Specialist (BSS) title — and the session we want on your calendar is the one led by Sebastian Lieb, a BSS graduate and structural-timber engineer who now runs computational workflows at Pirmin Jung, one of Switzerland’s serious wood-engineering offices. Lieb’s thesis, which we published as FOREST & HUMAN, did what most parametric demos never manage: it survived contact with a real engineering desk and kept running.

The idea underneath is older than the software. Graphic statics — the line-and-force-polygon method Karl Culmann formalised at ETH Zürich in the 1860s, and that Philippe Block’s group later rebuilt as an interactive computational discipline — always insisted that form and force are two readings of one drawing. What changed is that a single geometry model can now hold the architectural form, the load path and the numerical result at once. Lieb calls his result a “place of truth in Grasshopper”: one parametric model linking Archicad, Rhino/Grasshopper, RFEM (Dlubal’s structural solver) and Excel, so that a design change propagates instead of triggering a week of re-entry.

←TODAY: In 2026 a mid-size Swiss timber office still loses days re-keying geometry between the architect’s model, the structural solver, and the spreadsheet that owns the loads. →3012: By the Zurich-3012 horizon the model is the calculation — form, force and cost read from one source, argued in one room. Fulcrum: The saving is not the automation; it is that the same drawing finally answers to architect and engineer without either of them lying to the other.

Why now and not five years ago? The connective tissue matured — the data half of the live Rhino/Grasshopper↔Archicad link, not just the geometry, is now stable enough to carry a fully-informed structural model back into the BIM tool, load-transfer diagrams included. Lieb’s pipeline imports reference lines into Grasshopper, builds the parametric 3D structural model there, feeds loads straight from an Excel file into RFEM for calculation, returns the results to Grasshopper for processing, and exports the finished model to Archicad. Grasshopper is not a plugin in this story; it is the hub that always holds the latest data.

On a working engineer’s desk the point lands sharp. If you have ever watched a section change ripple through a project — new member sizes, new connection details, a fresh export, a re-import that half-works — you already know the tax this removes. It is why Lieb still uses a version of the tool at Pirmin Jung: a wood building is a system of load paths, and a parametric spine that keeps them honest through every revision is worth more than any single clever geometry. We have covered this thread across the BSS archive — Erick Allende’s climate-adaptive housing script, Rosalie’s canton-data buildings in Swiss-Landia, part of a cohort of 22 certified specialists — but Lieb’s is the one that closes the loop all the way to a certified structural result.

Atelier: For a Büro adopting AI and automation, the honest lesson from Lieb’s pipeline is that the win comes from wiring your existing tools together, not from buying a monster that replaces them — the same “puzzle pieces before monster tools” logic the BSS teaches. An office living with these workflows feels the difference on the third design change, not the first. Your Monday move: pick the one handover in your practice that already bleeds hours — most likely architect-model → structural-solver — and this week map every value that gets re-typed by hand across that gap. That list is your automation backlog, and it is the exact conversation the Lieb session is built around.

Hack: Read the axial force straight off a two-force member before RFEM ever opens, so you can sanity-check the solver instead of trusting it blind. Graphic statics says a member’s force is the geometry of the load projected onto the direction of the form. Five lines of Python at your desk turns that into a number you can defend in a meeting:

import numpy as np
a, b = np.array([0.0, 0.0]), np.array([3.0, 4.0])   # two-force member
u = (b - a) / np.linalg.norm(b - a)                 # unit vector along member
axial = float(np.dot(np.array([0.0, -12.0]), u))    # 12 kN load, projected
print(round(axial, 2))   # -9.6 kN -> compression

The sign is the whole lesson: negative is compression, positive is tension. If RFEM disagrees with your hand projection on a simple member, one of you has the geometry wrong — the model owes you an answer you could have estimated.

One trade-off, stated plainly: a place-of-truth pipeline concentrates risk as much as it concentrates data. When one Grasshopper definition owns the geometry, the loads and the export, a broken upstream reference can quietly corrupt everything downstream — so versioning and validating your own model becomes non-negotiable; the automation does not absolve you of reading the export. And it takes real time to build: the BSS runs five months at six hours a week, structured as three parts — Hello, Data, Party — before the final project, per the BSS Title page.

The September session is for architects and engineers who already model in Archicad or Revit and want the Grasshopper spine that connects design to structural truth; timber and structural specialists curious how a solver like RFEM or Dlubal folds into a live parametric loop; and offices weighing whether to build their own automation rather than rent it. For the Archicad side of that question — how you actually hold the AI-and-automation layer on top of your authoring tool, whether through GDL craft or the PAZ-BOX and PAZGPT add-ons — the PAZ Archicad add-ons are where we keep that work.

A school that only sells dies; a school that only remembers fossilises. We are re-telling Lieb’s graduation — you can watch the FOREST & HUMAN session in full — precisely so the next cohort inherits it and improves on it. Edition XIV starts 1 September 2026. Bring the one workflow in your office that hurts most, and come ready to wire it.