The flagship project of our AI-Powered Research practice is now live on this site: the Treatise of Physics, a full treatise published not as a static document but as a living, browsable book.
One source of truth, many faces
The treatise is written and maintained as a single TeX repository — the same kind of source a traditional physics text would use. From that one source, our publishing pipeline generates everything readers see:
- more than 5,000 interlinked HTML pages, so every definition, derivation, and result is a first-class page you can link to,
- full-text search across the entire treatise,
- the original figures, references, and bibliography, rendered faithfully from the TeX source,
- a downloadable PDF of the whole book for offline reading.
Nothing is edited by hand after export. When the source changes, the entire site is regenerated — which is what makes it a living book rather than a snapshot.
A knowledge graph you can walk through
The most distinctive part of the project is the interactive knowledge graph. Instead of a linear table of contents, the treatise exposes its own structure: concepts, the logical dependencies between them, and the sources they rest on, all navigable as a graph. You can follow a chain of reasoning step by step and see exactly what each result depends on — the kind of structural transparency we believe research publishing should offer by default.
Verified on every publish
Because the book is rebuilt from source on every release, quality has to be enforced by machines, not by proofreading. Every export runs through an automated integrity gate that checks the generated site — a single dead internal link is enough to stop a release from going out. The same pipeline publishes the treatise into the English, Spanish, and German sections of the site, so it is always reachable from wherever you are browsing.
Why we build this way
The Treatise of Physics is both a research project and a demonstration of what our AI-Powered Research methodology looks like in practice: a single verifiable source of truth, machine-checked structure, knowledge graphs instead of opaque prose, and a publishing loop that keeps the public artifact in lockstep with the underlying research.
Start reading the Treatise of Physics — and if this way of working with knowledge is something your organization could use, our AI-Powered Research page explains how we apply the same tooling to other domains.