Documents
Every way documents arrive, one pipeline.
Browser upload, a capture email address, a REST API and scheduled link collections all feed the same pipeline, so a document behaves the same no matter how it arrived.
Ingest
Four channels, zero setup for most.
Every organization gets a capture address. Developers connect systems with an API key and secret.
Ad-hoc files and first tests
Browser upload
Drag files onto the drop area. A floating progress panel tracks the transfer while you move to other pages.
Suppliers and colleagues
Capture email
Send or forward an email with attachments to your private address. Every supported attachment becomes a document.
Line-of-business systems
REST API
Push files as Base64, or register files already in your storage folder, up to 500 at a time.
Scheduled collection
Link collections
A batch runner visits supplier portals and file locations on a cron schedule and collects documents for you.
Hands-free ingestion
Set up a mail rule in Outlook or Gmail that forwards invoices from specific senders to your capture address.
Read
OCR and text extraction for 75+ formats.
Files are read with the OCR engine configured for their content type, and previewed right in the app.
Word & Excel
PowerPoint
OpenDocument
Images
E-books
Markup
Data files
Maximum size
200 MB per file.
Accepted
Only active content types. Other files are rejected as Not supported.
Duplicates
Same content (CRC32) and size as an existing document: skipped by default, or set to overwrite.
Prepare
Bundles split. Versions tracked.
Background tasks clean up what arrives before anything is extracted.
01 · Split
AI document splitting
Bundled PDFs, such as an order and its invoice, are separated into business documents. AI detects each type and extracts its content.
02 · Version
Automatic versioning
A re-uploaded file becomes a new version of the existing document instead of a duplicate.
03 · Compare
Version comparison
AI reports what was added, removed and changed between two versions. With auto compare on, it runs for every new version and can email the differences.
Extract
You define the fields. AI fills them in.
Each document type, such as General Invoice, owns a JSON schema and its own search index. AI maps the document content onto that schema, and the result becomes searchable field by field.
01
Waiting for pre processing
Stored, queued for the pipeline.
02
Indexed
Text is in the index: keyword and semantic search work.
03
Searchable
Fields are extracted: AI search works too.
// Document type: General Invoice
{
"invoiceNumber": "INV-2026-0814",
"issueDate": "2026-08-14",
"supplier": { "name": "Marble Systems" },
"totalAmount": 1840.00,
"lineItems": [
{ "description": "Pencil, HB", "quantity": 120 }
]
}
Start with 100 free credits.
Uploading, keyword and semantic search are free. You pay only for AI processing and storage.
Pricing
1 USD = 10 credits
Sign-in
Microsoft or email
Formats
75+
Max file size
200 MB