Most arguments scatter across papers, posts, and threads — and evaporate. A claim tree gives them a stable structure to gather against, the way a cathedral gathers centuries of work into a single, standing thing.
A hierarchical decomposition of one top-level claim into sub-claims. No scoring, no attribution — just structure. The part that stays.
Any external text — a paper, a blog post, a comment thread. Raw material that already lives out in the world.
Fragments pulled from a document and mapped onto specific nodes, with a link home. People and AI propose the mapping; you confirm it.
Relatable, genuinely contested, and it decomposes cleanly. So we break it apart — not to answer it yet, but to give every future answer a place to live.
We break the question into the dimensions that actually decide the answer — the places where reasonable people disagree.
For whom? For what tasks? Over what horizon? Under what conditions? Each branch splits again into the distinctions that evidence will eventually have to land on.
No scores. No verdicts. Just the shape of the disagreement — the structure everything else will accumulate against.
People and AI pull fragments from each new source and attach them to the nodes they bear on. Here is the first one landing — the same scaffold, now catching evidence.
We ran a nine-month randomised experiment at Ctrip, a 16,000-employee travel agency, assigning a group of call-centre staff to work from home while their colleagues stayed in the office.
Home working led to a 13% improvement in performance, of which roughly nine points came from working more minutes per shift — fewer breaks and sick days — and four from handling more calls per minute in a quieter, more convenient environment.
Home workers reported higher satisfaction, and attrition fell by half. Yet when later given the choice, over half asked to return to the office — and the gains for that group narrowed once they did.
The setting is a specific one: routine, individually-measured call-centre work, where output is unusually easy to quantify. Whether the same holds for collaborative or creative roles remains open.
Each new source is processed against the same scaffold. Evidence accumulates; the structure sharpens. This is how a question becomes a cathedral — one annotation at a time.