Introducing s-1: Sievly’s new document reading model
All posts
Engineering
Reading tables that span pages
A table that continues on the next page is still one table. This post covers how s-1 stitches multi-page tables back together.
Lena Varga
·
Research Engineer
sievly
Tables across pages
Financial statements, lab reports and policy schedules often run a single table across several pages. Treated page by page, each fragment loses its header and the totals stop making sense.
Detecting continuation
s-1 looks for repeated headers, matching column positions and running totals to decide whether a table on one page continues the table on the last.
Rebuilding the table
Rows from each page are merged under a single header and returned as one table, with a citation for every cell that points to its original page.
Edge cases we still watch
Rotated continuation pages and tables that change column order mid-document are flagged for review rather than merged silently.
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.