Introducing s-1: Sievly’s new document reading model
Parse
Turn any document into clean, structured content
Layout-aware chunks, tables and figures in reading order, with a citation on every block. One call for any file type.
POST /parse
Helping everyone from early-stage startups to global enterprises unlock their data.
Quarryline
Brindlewood
nimbleton
brightwick
vantrel
PLINTH & CO
OAKMERE
ARROWFEN
Halvory
pebblegate
Mirrowind
quillstack
Parse
Read documents the way people do
DEFINITION
Parse finds the layout of every page, keeps tables and figures whole, follows reading order and returns structured content you can search, chunk or pass to a model.
WHO IT’S FOR
AI engineers building retrieval and assistants, data teams ingesting large archives, and product teams adding document features.
THE PROBLEM IT SOLVES
Plain OCR flattens pages into text. Headers drift from their rows, footnotes lose their numbers and charts disappear. Parse keeps the structure that makes content true.
Parse in the platform
Using Parse in your pipeline
Parse is usually the first step. Everything downstream improves when the input is clean.
#1
When you are building retrieval
Get layout-aware chunks that keep tables, captions and headings together, ready to embed.
#2
When documents are messy
Scans, rotated pages, handwriting and nested tables come back as structure instead of noise.
#3
When you need to show sources
Every block carries page numbers and bounding boxes, so answers link back to the page.
Where teams use Parse
Ingestion that holds up on real files
Common jobs where Parse replaces a stack of tools.
RAG ingestion
Turn archives of PDFs into cited chunks for your vector store.
Table reconstruction
Rebuild tables that span pages, with merged cells and headers intact.
Scanned documents
Read low-quality scans and faxes with automatic page rotation.
Figure summaries
Describe charts and diagrams so they become searchable.
Multilingual files
Read documents that mix languages and scripts on one page.
Spreadsheets and slides
Bring spreadsheets and decks into the same structured output.
Parse your own documents in Workbench.
Why Parse
Why teams use Parse
01
One call for every file type
PDFs, scans, images, spreadsheets and slides go through the same endpoint.
02
Structure you can trust
Tables, lists and figures stay whole and in reading order.
03
Citations by default
Page numbers and bounding boxes on every block.
04
Tunable chunking
Choose chunk size and strategy to match your retrieval setup.
How Parse works
How Parse works in four steps
■ STEP 01
Install the SDK
Add the Python or Node client to your project.
pip install sievly
■ STEP 02
Send a file
Upload a file or pass a URL to Parse.
client.parse(file)
■ STEP 03
Choose options
Pick chunking, figure summaries and output format.
chunking: layout
■ STEP 04
Use the output
Store chunks, render tables or pass content to a model.
→ chunks.json
Built for production
Built for production workloads
Parse runs on the same infrastructure as every Sievly endpoint, with the controls enterprise teams expect.
Built for regulated teams
Security controls cover the whole platform, with agreements available on Growth and Enterprise.
Deploy in your environment
Run Sievly in your own cloud when data cannot leave your network.
High-availability uptime
Built for steady throughput at any volume, with SLAs on Enterprise.
The rest of the platform
The endpoints around Parse
/extract
Extract
Pull the fields you define, with a citation on every value.
Read more →
/split
Split
Break long, multi-document files into logical parts.
Read more →
/classify
Classify
Label and route documents by type before they move on.
Read more →
/edit
Edit
Fill forms and update documents from plain language.
Read more →
/workbench
Workbench
Design, test and deploy pipelines without code.
Read more →
Further reading