Anthropic measured AI occupation by occupation. For engineering and architecture, the verdict is brutal — and it's good news.
On the office-based tasks of engineering and architecture, AI could already handle 82% of the work. Today, it does 18% of it.
That gap is one of the largest in the entire economy. And it's exactly where the next decade of construction will be decided.
From a study by Anthropic (the company behind Claude), « Labor market impacts of AI », 2026. Millions of real usage data points cross-referenced with the theoretical potential of AI, occupation by occupation, task by task. Not a consulting-firm projection: observed usage.
Construction isn't one block.
The conclusion is simple: the opportunity isn't on the job site, it's in your offices. Reading a 150-page technical specification (CCTP), building a technical bid, analyzing tenders, drafting meeting minutes — that's precisely what AI does well, and what your teams still do by hand.
Since ChatGPT: −14% in hires among 22-to-25-year-olds in the most exposed occupations (same study). No mass layoffs — but more is being done with fewer new recruits.
For a junior engineer set to work in the office, mastering AI is no longer a "nice-to-have." It's a condition of employability. For the company training its teams now, it's a head start.
BIM went from "gadget" to "required in tenders" in just a few years. Those who got on board early captured the market. Office AI is following the same curve — only faster.
The majors set up committees, charters, POCs that drag on for 18 months. A company of 50 to 200 people can put a useful practice in place within a few weeks. Agility beats resources. This is the moment when an SME can overtake a large group on this front.
AI won't pour your concrete. But it can produce your documents and your drawings in a few minutes — while your teams focus on what matters.
Live, on real documents. Then 30 min to talk it through on yours.