Introducing s-1: Sievly’s new document reading model
All posts
Cookbooks
Cookbook: build a cited RAG ingestion pipeline
Use Parse to turn a folder of PDFs into layout-aware chunks with citations, ready for your vector store.
Priya Nair
·
Developer Advocate
sievly
Cited RAG ingestion
This recipe takes a folder of mixed PDFs and produces chunks your retrieval system can cite back to the page.
1. Parse each file
Call Parse with layout chunking turned on. You get chunks that follow headings and tables, each with page numbers and bounding boxes.
2. Add figure summaries
Turn on figure summaries so charts and diagrams become searchable text instead of blank space.
3. Embed and store
Embed each chunk with your model of choice and keep the citation metadata next to the vector.
4. Answer with sources
When your assistant answers, return the citations with it so users can open the source page.
More from the blog
sievly
Introducing s-1
Announcements
Introducing s-1, our new document reading model
s-1 reads PDFs, scans and spreadsheets the way a careful analyst does, and returns structured data with a citation for every value. It is available in preview today.
sievly
Reading order
Research
Research note: reading order in complex layouts
Multi-column pages, sidebars and callouts confuse simple reading order. We share what we learned teaching s-1 to follow the page the way a person would.
sievly
Quarryline × Sievly
Case Studies
How Quarryline turns research notes into a searchable knowledge base
Quarryline’s data team uses Parse and Extract to turn thousands of analyst notes, filings and spreadsheets into clean, cited records their AI assistant can trust.