Ovrin’s architecture is built around three key seams: Model, OCR, and Renderer.
Model
The model seam produces structured JSON from the document content. It is the part that turns document text and instructions into structured data.
OCR
OCR recovers text and layout from rasterized pages. This is used when page content is not available natively.
Renderer
The renderer turns a document page into an image suitable for OCR or visual analysis.
Why these seams are important
This separation keeps the core free from provider-specific dependencies and lets you swap adapters without rewriting your extraction logic.