OSDIS

OSDIS is a document-understanding substrate built on a simple idea: make documents machine-readable without making them worse for people. It adds a shadow representation to a PDF — a layer carrying the structure a machine needs, while the document still reads exactly as intended to a human.

The PDF is everywhere and machine-hostile: it preserves how a page looks and discards what it means, so every system that consumes one has to reverse-engineer its structure, badly. The fix is not better guessing after the fact — it is putting the structure in before the document is frozen.

Arc holds OSDIS in the document-intelligence family alongside Hunter-RAG, as enabling technology and a candidate standard — a representation layer that, if it spreads, makes documents legible to machines at the source. The novelty here is protected.

Prototype