Form-finding is the oldest new idea in Swiss structural design. Hang a chain, let gravity draw the pure funicular line, invert it, and you have a shape that carries load in compression alone — the method Heinz Isler poured into his shell roofs and that the Block Research Group at ETH Zurich has since turned into computational graphic statics anyone can script. What we love about it is the honesty: the geometry is the calculation. This week we want to point that same lens at a working desk, because two engineers are about to show what happens when a practitioner stops treating form-finding as theory and writes the tool themselves.

On 1 September 2026 — the same morning our BSS Title Edition XIV opens — we host Camilla Seidler and Peter Steiner, both engineers at Pirmin Jung, the Swiss office that has made timber engineering its whole identity. This is a PAZ-Next talk, and the reason we scheduled it is simple: they each built software their profession needed and could not buy.

Camilla is developing a tool that computes structural loads for wood structures using form-finding — reading force directly off geometry rather than fighting a spreadsheet into agreement with a model. Peter went after window and façade coefficients, where he found the existing Swiss software solutions dated and inflexible, so he wrote his own to analyse any building-system façade against the window geometry coming out of the architect’s design. Neither started as a programmer. Both are finishing the BSS Title and shipping working tools as they go.

←TODAY: A Pirmin Jung engineer form-finds timber loads in a script she wrote, because the off-the-shelf Swiss tool could not bend. →3012: The office no longer buys its structural logic — it grows it, one desk-built tool at a time. Fulcrum: The moment an engineer can read force off geometry, the tool becomes a service, and the service becomes the firm’s moat.

This is the thread PAZ has been pulling for years. In our archive piece Forest & Human, BSS graduate Sebastian Lieb built exactly this kind of “place of truth”: a Grasshopper hub linking ArchiCAD, RFEM and Excel so loads flow from spreadsheet to solver to BIM model with no manual re-entry — a pipeline he still runs at Pirmin Jung today. Seidler and Steiner are the next turn of that wheel. The BSS Title is a five-month, six-hour-a-week mentorship, and its whole pedagogy is that you learn the puzzle pieces before you touch the monster tools — so that the tool you eventually reach for is one you understand well enough to rebuild. A prior group graduated 38 certified Building System Specialists, and every student inherits over 15 years of PAZ scripts and templates as a running start.

Why this matters at your desk: Switzerland’s timber sector is scaling fast, and the analysis tools have not kept pace. When your façade solver is a decade out of date, every non-standard window becomes an afternoon of manual coefficient lookup. The engineers who close that gap in-house are the ones whose services compound — Pirmin Jung expects tools like these to save real money over the next five years precisely because they turn a recurring manual cost into a reusable asset.

Hack: Turn a distributed area load into the line load your structural solver actually wants, in three honest lines you can paste beside a Grasshopper Python node. The trick every desk-built tool starts from is that RFEM or Dlubal eats line loads, but the architect hands you area loads and a spacing — so you multiply once and keep the geometry as your source of truth.

import numpy as np
stations = np.linspace(0, 8.0, 9)   # timber beam: 8 m span, 1 m stations
q_area   = 2.0                      # kN/m2 characteristic area load
spacing  = 1.2                      # m, joist tributary width
line_load = q_area * spacing        # kN/m -> feed straight to RFEM/Dlubal

That one multiplication is the seed Camilla’s form-finding tool grows from: once the load lives in code next to the geometry, you can vary the span, re-solve, and never re-type a number again.

Atelier: For an office weighing whether to let engineers build their own tools instead of buying another licence, the honest trade-off is maintenance — a script one person wrote is a script one person has to keep alive, and that lock-in is real. The Monday move: pick the single most out-of-date recurring calculation on your desk — the façade coefficient lookup, the manual load transfer — and give one engineer a half-day to prototype it in Grasshopper Python, the same Archicad-automation muscle our workshops build, before committing to any new software purchase.

Come listen to two people who did exactly that. The talk is our host’s invitation, not a sales pitch — a school that only sells dies, and a school that only remembers fossilises, so we bring you the practitioners while their tools are still warm. Bring the one calculation your office keeps doing by hand, and watch how Seidler and Steiner turned that same friction into software.