■ SAFE RISK ■ Construction & Trades
A miyadaiku reads timber and site conditions that no CNC program accounts for. Robots can cut the joints; they cannot decide which face of which tree goes where, or how a 300-year-old temple frame has moved.
“No nails, no screws. Just joinery perfected over 1,400 years. Good luck programming that.”
Our AI replacement risk score — how we score jobs
Traditional Japanese carpentry is structural engineering carried out with hand tools and a memorised joinery vocabulary. Selecting timber by grain, growth direction and how the tree stood on its slope, then orienting each piece so the building ages the way it should. Sharpening and tuning kanna planes and chisels — a skill that consumes years by itself. Laying out with sumitsubo and square, cutting interlocking joints that lock without fasteners, and raising frames that flex under seismic load rather than resisting it. Restoration work adds forensic reading of an existing structure and matching centuries-old techniques.
CNC has genuinely entered the field. Automated joinery machines cut mortise-and-tenon and complex compound joints for timber-frame construction at speed, and prefabricated post-and-beam housing in Japan uses this heavily. 3D scanning documents existing structures for restoration, BIM models the frame before a piece of wood is touched, and generative tools handle layout permutations. For standard residential timber framing, the machine is faster, cheaper and consistent — that battle is settled.
The high craft resists on three fronts. Material variability: each timber is unique, and the decision of where it goes is aesthetic and structural at once. Site conditions: restoration means working with a building that has settled, twisted and been repaired badly at some point, where no drawing is accurate. And the cultural mandate: shrine and temple work is intentionally done by hand because the transmission of the technique is part of the point. Our risk score of 20 reflects mass housing lost to automation long ago, while heritage and bespoke work stays secure — constrained by apprentice numbers rather than machines.
Automatability: our editorial assessment of current and near-term AI capability
The commodity end automated decades ago and CNC joinery is now standard in Japanese residential framing. Through the 2030s expect scanning and BIM to become routine in restoration planning too. Heritage carpentry itself stays hand-worked past 2040 by design as much as necessity. The genuine threat is demographic — a shrinking pool of apprentices willing to spend a decade learning before earning properly.
Many joints, yes — automated joinery machines handle mortise-and-tenon and complex compound cuts accurately and fast, and Japanese prefabricated housing relies on them. What machines cannot do is choose which timber goes where based on grain and how the tree grew, or adapt to a historic frame that has twisted over three centuries.
In heritage restoration and high-end bespoke work, yes, and skilled practitioners are scarce. In ordinary residential framing, the work is machine-cut and assembled, which is a different trade. The barrier is the apprenticeship: years of low pay learning tool maintenance and layout before you are trusted with structural work.
Partly because historic structures demand it — every joint must match existing conditions that no drawing captures — and partly because the transmission of technique is itself considered part of the heritage being preserved. Cultural property rules and the periodic rebuilding traditions exist specifically to keep the knowledge alive in living hands.
Let machines do rough dimensioning and repetitive cuts, and use scanning and BIM for survey and planning, which clients increasingly expect. Keep hand skills for the joints and finishes that define quality. The carpenters winning restoration contracts today typically offer digital documentation alongside traditional technique, not one instead of the other.