Native PDF text before OCR
Why a text layer should be the first extraction path for modern Gazette documents.
/ BLOG
How we make the
public record useful.
Product notes about the evidence layer, search interface, and the choices behind GazetteIntel. Specifics over slogans.
ALL NOTESPRODUCT · RESEARCH · BUILD
The Gazette problem is not the PDF.
We started by looking at the archive. The faster idea was not another portal. It was a question, a source, and a watch that keeps working after you leave.
Read the note ↗What makes an answer citable?
Search, page boundaries, and the small details that keep a source usable.
TOOLS
The product should feel like a conversation, not a filing cabinet.
Explore the planned question pin, alert rule, collection, and API surfaces.
View tools ↗