Two distinct latent assets

Two distinct latent assets. (1) A labelled multilingual trade-document corpus: every production extraction is an image-to-structured-JSON pair on real SEA bills of lading, customs declarations, commercial invoices, packing lists, certificates of origin, phytosanitary certificates, delivery orders and proof-of-delivery scans — across Thai, Vietnamese, Bahasa Indonesia/Malay, Tagalog, Khmer, Chinese and Japanese, across 50+ carrier layouts, with field-level ground truth confirmed by downstream customs submission. (2) A visual-forensics corpus: forged/tampered versus genuine document images with splicing, copy-move, retouching and AI-generated/edit labels at suspicious-region coordinate level, plus document version-pairs with semantic-diff annotations.

Open buyer room →

Sample

Brickroad customers

Sample this dataset before you buy it.

Your sourcing agent asks Turbolens and files the sample in your catalog — private to you.

Start 7-day trialNothing is charged until the trial ends.

Coverage

Universe, instruments and categories.

Industry: Application Software
Instruments: equities
Categories: AlternativeSupply Chain

Dataset card

Type: not stated
Format: Multimodal — document images and digital PDFs paired with structured JSON extractions (document_type, language, fields, line_items), plus forensic outputs carrying verdicts, probability scores, suspicious-region coordinates and heatmap references
Volume: ~0.5-5 million processed document pages cumulatively (est. — freemium daily scan allowances across a low-thousands user base plus enterprise API throughput on shipment volumes, accumulated only since a ~2024 launch; the vendor publishes no processing totals, so this is an order-of-magnitude bound and should be sized against their actual logs in diligence)
Users: ~1,000-10,000 registered users on the free/trial tier and low hundreds of paying accounts (inferred from a freemium ladder — free limited daily scans, $49.90/month premium, custom enterprise API — plus launch-platform posts as the main acquisition channel)
History: 3 years
Update frequency: not stated
Growth: Active — successive dated launches (2024 launch, 2025 extraction/parsing additions, a March 2026 major launch) with synthetic-media screening still flagged as early-access, and an expanding vertical surface (banking, insurance, healthcare, logistics, government) each carrying named regulator and core-system integrations
Launched: ~2024 (DocumentLens first launched 2024, per launch-platform history; capability additions announced 2025)
Delivery: not stated
Entity mapping: not stated
Sample: not stated
Point-in-time: not stated
Licence: not stated

supplier index read from their site