■ SAFE RISK ■ Construction & Trades
Safe, and quietly short of people. Conservation of historic glazing is bespoke, irreversible and legally supervised work on unique objects — the least attractive target imaginable for automation.
“800-year-old cathedral windows need hands that understand medieval glass. AI can't even.”
Our AI replacement risk score — how we score jobs
The work starts long before any glass is touched. Condition surveys documenting every crack, paint loss and previous repair; rubbings and photographic records so the panel can be rebuilt exactly; scaffolding and careful removal of leaded lights that may weigh a great deal and be structurally tired. Then the bench work: dismantling old lead cames, cleaning corrosion crusts off medieval glass without stripping the surviving paint, edge-bonding fractures with conservation resins, matching replacement glass by colour, texture and refractive quality, re-leading, cementing, and installing protective glazing with the right ventilation gap so condensation does not accelerate decay.
Digital tools have genuinely improved the documentation half. High-resolution photogrammetry and rectified imaging produce accurate records; CAD reproduces cartoons and lead lines; spectral imaging can reveal lost paint or earlier schemes; databases track panel-by-panel condition across a whole cathedral programme. Waterjet and CNC cutting can produce accurate shapes for new commissions, and glass painting can be reproduced by digital printing for replica work. Grant applications, condition reports and heritage-consent paperwork are all becoming faster to produce with AI assistance.
The hands-on core does not move. Each panel is a unique assembly of degraded, chemically unstable material where an aggressive clean destroys the object permanently. Judging how far to intervene is an ethical decision governed by conservation charters and heritage authorities, not an optimisation. Matching a 13th-century pot-metal blue involves eye, experience and hand-blown stock. Working at height in a cold church, easing a bowed panel out of its groove, is not a robotics problem anyone will fund. Our risk score of 22 mostly reflects admin and documentation automation, plus competition in the cheaper replica end of the market.
Automatability: our editorial assessment of current and near-term AI capability
No meaningful disruption on the horizon. Documentation, imaging and reporting will keep getting faster through the 2030s, which mostly frees bench time. The replica and new-commission market will see more CNC cutting and digital printing. Conservation itself stays a hand craft well past 2040 — its binding constraint is the number of trained conservators and the size of heritage budgets, both of which are already tight.
Not realistically. Every panel is a unique, chemically degraded object where an error is permanent, and the decisions are ethical as much as technical — how much to intervene, what to leave as honest evidence of age. Robotics has nothing to offer for easing a bowed medieval panel out of a mullion forty feet up a scaffold.
Demand is steady to rising, because historic glazing keeps ageing and protective glazing programmes are long-term commitments. Supply of trained conservators is the problem. The limiting factors are heritage funding cycles and apprenticeship pipelines, not competition from technology, which makes it one of the safer craft careers available.
Documentation most of all: photogrammetry, rectified photography and spectral imaging have transformed condition recording and made panel-by-panel tracking across large programmes practical. CAD and waterjet cutting help with new commissions and replica work. Reporting and funding applications are the obvious place AI writing tools save hours each week.
Usually through a conservation or glass programme followed by studio apprenticeship, since accreditation bodies expect supervised practice. Building a portfolio of documented projects matters more than credentials alone. If you can combine bench skill with digital survey competence and grant writing, you will be unusually employable in a field where most studios are small.