Document intelligence API
Turn documents, audio and video into data your systems can use.
Send any file. Get back accurate, structured text with the layout intact, checked by an AI quality-control pass, and delivered wherever your applications need it.
4
Input types
Documents, images, audio, video
$0.012
Per page
Quality control included
4
Vector destinations
Pinecone, Qdrant, Weaviate, pgvector
$0
Platform fee
No subscription, no seats
The problem
Most of your useful information is locked inside files.
Invoices, contracts, scans, recorded calls and screen captures all carry information your business runs on. Getting it out reliably is where projects stall.
Manual rekeying
People retype what a machine should have read, which is slow and quietly introduces errors into records.
Lost structure
Generic extraction returns a wall of text. Tables collapse, columns merge, and the meaning goes with them.
Too many tools
One vendor for OCR, another for transcription, another for embeddings — and glue code holding it together.
How it works
From raw file to usable data.
Four stages run on every job. Select a stage to see what happens inside it.
Step 3 of 4
Verify
An AI quality-control pass reviews the extraction and corrects likely recognition errors before you see it.
Capabilities
Everything the job needs, in one place.
One API for every format
Documents, images, audio and video go through the same endpoint. No separate vendor for transcription and no template setup per document type.
Quality control on every file
An AI pass reviews the extraction and corrects likely recognition errors — misread digits, broken dates, confused characters — before the result reaches you.
Structure that survives
Tables stay tables, columns keep their reading order, and every block comes back with its position on the page so results can be traced to the source.
Ready for search and AI
Embeddings can be written straight into Pinecone, Qdrant, Weaviate or pgvector, so retrieval and AI features work the moment a file finishes.
File types
Send the file you already have.
Documents, images, audio and video are all supported through the same endpoint.
Documents
Multi-page files with tables, columns, forms and stamps. Reading order and layout are preserved.
Billed per page
Images and scans
Photos, scans and screenshots, including skewed, low-light and handwritten captures.
Billed per page
Audio
Calls, meetings, interviews and voice notes turned into timestamped, searchable text.
Billed per minute
Video
On-screen text from slides, screen recordings and footage, aligned with the spoken track.
Billed per minute
AI search and assistants
Ready to search and answer questions.
Finished files are split into searchable chunks that keep their file, page and section labels, then embedded and written into your own database. Search, chat and assistants work without a second pipeline.
Total due SGD 12,480.00. Payment terms: net 30 days from invoice date…
Line items: Document processing services, quantity 1, amount 9,980.00…
Remittance details: account ending 4471, SWIFT reference NSCOSGSG…
Solutions by industry
Built for document-heavy work.
Financial services
Read every statement, invoice and filing without a rekeying team.
Read moreLegal and compliance
Make contracts, filings and discovery bundles searchable.
Read moreLogistics and trade
Clear shipments without waiting on manual document entry.
Read moreHealthcare and insurance
Turn intake forms, reports and claims into usable records.
Read moreAI and data teams
Ship retrieval over your own files without building a parsing stack.
Read morePricing
One rate. Everything included.
No subscription, no seats and no minimum. Top up a wallet and pay only for what you process. Your first top-up receives $5 in bonus credit.
$0.012
per page
Documents and images
$0.03
per minute
Audio
$0.06
per minute
Video
Put your first file through the pipeline
See accurate, inspectable results before you commit.
FAQ