# A rebuilt parser for PDF steps

> <p>Steps that read a PDF (Extract from PDF, Pull from file queue, and Extract from email) now run on a rebuilt parser. It uses a newer model, so it holds up better on messy documents and awkward table layouts, and it reads the document as markdown first, which makes larger PDFs parse faster and lets it handle documents that used to be too big to get through. Steps you already built keep the parser they were created with, so add a new step to pick up the improvements.</p>

Source: https://parabola.io/updates/a-rebuilt-parser-for-pdf-steps