10-K financial extraction
Pull figures and disclosures from long annual filings, each with a citation back to the page it came from.
Docen LayoutDocen Extract
THE CHALLENGE
A 10-K can run well past a hundred pages of dense tables, footnotes, and narrative. The number you need is in there — the trick is finding it and proving where it came from.
THE DOCEN APPROACH
Docen segments the filing, preserves its tables as structured cells, and extracts the metrics and disclosures you define, with citations that make every figure auditable.
- Segment the filing into items and sections.
- Keep financial tables intact as structured cells.
- Extract metrics and disclosures against your schema.
- Cite the page and span for every value.
extract → filing.json
{
"fiscal_year": 2025,
"total_revenue": "$4.82B",
"operating_income": "$0.91B",
"cash_and_equivalents": "$1.34B",
"risk_factors": 27,
"citations": { "total_revenue": { "page": 54, "item": "8" } }
}100+
pages per filing
100%
figures cited
auditable
by design
[→]GET STARTED
Try it on yourown documents.
Bring a sample of the documents behind this use case and we'll show you the structured output.