■ HIGH RISK ■ Construction & Trades
No one is building a thatching robot for a trade with a few thousand practitioners worldwide — the market is too small to automate. The risk to thatchers isn't AI; it's the slow economics of a heritage craft, which is a different kind of replacement.
“There are so few of you left that AI didn't bother training on your job.”
Our AI replacement risk score — how we score jobs
Thatching is roofing with harvested plant material — water reed, combed wheat straw, or long straw — layered, fixed with hazel spars and steel crooks, and dressed into a weathertight coat that can last decades. A thatcher's day is physical and meticulous: stripping old coatwork, sorting bundles by quality, laying courses to a consistent pitch, shaping the ridge, and reading how water will actually shed off this particular roof's valleys and dormers. Most work in the UK and parts of Europe is on listed heritage buildings where the material and method are legally prescribed.
Automation has almost nothing to grab here. The work happens on irregular old roof timbers, every building is unique, the material is organic and variable bundle to bundle, and the total market is microscopic — nobody funds robotics R&D to chase a few thousand jobs. Where technology does touch the trade, it's mundane and helpful: drones for roof surveys, better moisture meters, online scheduling, and mechanized reed harvesting upstream in the supply chain. AI could plausibly help with quoting and condition assessment from photos, but that assists the thatcher rather than replacing one.
So why does our risk score sit at 51 rather than 10? Because the threat is economic, not algorithmic. Thatch competes with cheaper modern roofing wherever heritage rules don't mandate it, skilled labor costs keep rising, insurance premiums on thatched properties climb, and imported reed supply is volatile. The craft survives on conservation regulation and aesthetic demand — both real, neither growing fast. The trade is safe from machines and perpetually exposed to markets, apprenticeship shortages, and the slow attrition of the building stock that needs it.
Automatability: our editorial assessment of current and near-term AI capability
There is no automation timeline because there is no automation program — the market is too small and the work too irregular to attract one. The pressures that matter operate on craft-economy time: heritage funding cycles, insurance costs, reed supply, and whether enough apprentices enter to replace retiring masters. Expect the trade to look much the same in 2040, just contending with economics rather than robots.
About as safe as it gets. The work is performed on irregular historic structures with variable organic materials, and the global market is a few thousand practitioners — far too small to justify robotics development. The real risks are economic: heritage funding, insurance costs, material supply, and attracting apprentices.
Because our risk score measures livelihood pressure, not just robot proximity. Thatch competes with cheap modern roofing outside legally protected buildings, insurance premiums push some owners away, and the building stock needing thatch slowly shrinks. The craft won't be automated — but it can still contract.
In regions with protected thatched building stock — southern England, parts of Ireland, the Netherlands, Denmark — yes, and waiting lists are common because practitioners are scarce. Rethatching cycles and ridge repairs generate steady repeat work. The constraint is geography: demand is dense in heritage areas and nearly zero elsewhere.
Through apprenticeship with a working master, typically several years, sometimes supported by heritage-craft training schemes. Accreditation through recognized thatching associations matters for insurance-backed and listed-building work. It's a genuine commitment — but scarcity works in your favor once qualified, which few modern careers can claim.