■ MODERATE RISK ■ Creative Arts
No. AI can sketch a rose window in seconds, but it can't cut glass, solder lead, or convince a restoration committee. A craft that outlived the printing press, photography, and IKEA will digest image generators as one more design tool.
“Your craft survived 800 years. AI-designed patterns are just suggestions to your lead came.”
Our AI replacement risk score — how we score jobs
The work divides into design and fabrication, and fabrication is most of the hours. An artist consults with a client — a church, a homeowner, a conservation body — produces a design and a full-size cartoon, selects sheet glass for color and texture, then cuts each piece by hand, grinds edges, paints and kiln-fires details like faces and drapery, and assembles everything in lead came or copper foil before cementing and waterproofing the panel. Restoration work adds another discipline: documenting historic windows, matching century-old glass, and re-leading panels without losing original material. Installation means scaffolding, templates, and weatherproofing on-site.
AI has genuinely entered the front of that pipeline. Image generators produce design concepts and clients increasingly arrive with them; software already handles pattern drafting, glass-cutting optimization, and color visualization. CNC and waterjet cutting can profile glass for production work, and cheap printed film 'stained glass' has long since taken the bottom of the decorative market. For a studio, the honest effect is compressed design time and clients who think the design was the hard part.
It wasn't. Glass selection alone — reading how a sheet of mouth-blown antique glass transmits light at different hours — is judgment no renderer replicates, and painting and firing traditions like silver stain behave chemically, not digitally. Restoration, which anchors many studios' income, is governed by conservation standards that mandate documented hand methods on protected buildings. The craft's real vulnerabilities are economic: shrinking church construction and the patience required to train. Our risk score stays low because demand is niche but the moat around fabrication is deep.
Automatability: our editorial assessment of current and near-term AI capability
Design generation and pattern software are transforming the studio front-office now, and clients bringing AI concepts is already normal. Through the 2030s expect CNC cutting to absorb more production work while hand fabrication, painting, and restoration remain manual. By ~2040 the profession is smaller and more restoration-weighted — driven by church economics more than automation — with the surviving studios busier than ever on heritage contracts.
AI can design them — image generators produce convincing concepts in seconds, and clients increasingly bring them to studios. But a design is a picture; a window is cut glass, fired paint, and soldered lead built to survive weather for a century. Fabrication remains skilled handwork, so AI has changed the sketchbook, not the bench.
It's a shrinking-but-stubborn one. New church construction has declined for decades, which squeezed the traditional commission base — that's the real pressure, not technology. Meanwhile restoration of historic windows provides steady, protected work, and residential and architectural commissions have grown. Studios that blend heritage skills with modern marketing are doing fine; the craft has outlived worse.
They already handle repetitive production cutting, and that will expand. But most studio work involves one-off pieces in irregular, handmade glass where sheet selection and cutting judgment matter, plus painting, firing, and leading that machines don't do. Restoration standards on protected buildings often require traditional hand methods outright, keeping the core craft mandatory.
Adopt them shamelessly for concepting and client visualization — they compress the slowest conversations in the commission process. Then differentiate on what they can't touch: glass selection, painted detail, restoration expertise, and the credibility to work on heritage buildings. The artists at risk are those who only sold designs; makers keep their moat.