Scrapers have always been notoriously brittle and prone to breaking completely when pages make even the smallest of structural changes.

Scraping with LLMs bypasses that pitfall because it's more of a summarization task on the whole document, rather than working specifically on a hard-coded document structure to extract specific data.