Aekam Trade DocAI Trade Document & Fraud Intelligence
DocAI Engine v2.4 DEMO · SYNTHETIC DATA
Trade Finance · Pre-disbursement Screening

Catch the fraud
before the money moves.

Upload the trade documents behind a financing request. The DocAI engine reads every page, cross-checks the trade chain, hunts for tampering and duplicate financing, and returns a single Trade Authenticity Score with a clear disposition — in seconds, not days.

OCR field extraction Cross-document matching ELA / FFT tampering forensics SHA-256 duplicate-financing check Price anomaly screen
Built for commodity traders, trade-finance banks & insurers.
Financing request PENDING REVIEW
Applicant (Importer)Asia Pacific Resources Pte Ltd
Beneficiary (Exporter)Meridian Commodities DMCC
CommodityHMS 1&2 Steel Scrap
InstrumentImport L/C at sight
L/C referenceLC-SG-2026-44120
Amount requestedUSD 2,400,000
Reviewing deskDBS Bank — Trade Finance
3 documents presented for negotiation. The desk must decide whether to disburse, hold, or escalate.
Step 1 · Document intake

Ingest the presentation

These three documents were submitted in support of the financing request. Load them into the engine to begin analysis.

📄
Commercial Invoice
MC-INV-2026-0847.pdf · 1 page · 248 KB
Ready to ingest
Preview
🚢
Bill of Lading
BL-OHJEA260418.pdf · 1 page · 312 KB
Ready to ingest
Preview
🏭
Warehouse Receipt
WR-JBW-2026-3318.pdf · 1 page · 201 KB
Ready to ingest
Preview
Step 2 · OCR & field extraction

Reading every document

The engine classifies each document, then extracts structured fields with per-field confidence. No manual keying.

Step 3 · Cross-document comparison

One trade, three stories

Every field is reconciled across the invoice, the carrier's bill of lading and the warehouse receipt. Agreement is expected — divergence is the signal.

Step 4 · Forensic checks

Beyond the numbers

Mismatches can be honest errors. These three checks separate clerical noise from intent: tampering forensics, price anomaly, and duplicate-financing detection.

🔬 Document tampering — Warehouse Receipt High

Error-Level Analysis (ELA) and FFT residual scan localise where the document's compression fingerprint breaks. The Quantity Stored field was re-saved separately from the rest of the page.

Jurong Bonded Logistics — Warehouse Receipt JBW-WR-2026-3318
Depositor: Asia Pacific Resources Pte Ltd
Commodity: HMS 1&2 Steel Scrap
Quantity Stored: 5,000.000 MT
Received: 24–25 Apr 2026 · ex MV Ocean Harmony
ELA Δ +312% · re-compressed region
PDF producerAdobe Photoshop 25.x (image layer)
Page text layermixed — 1 field rasterised
Created25 Apr 2026 09:14
Last modified25 Apr 2026 18:47 (post-issuance)
Font on quantityCourier — differs from body (Helvetica)
Forgery likelihood
0%

📈 Price anomaly Medium

Invoiced unit price is screened against the 30-day market reference for HMS 1&2 (CFR SE Asia). A material premium with no contractual basis is a classic over-invoicing / TBML signal.

Invoiced: USD 480.00 / MT
Market reference: USD 400.00 / MT (30-day avg)
Premium: +USD 80.00 / MT  (+20.0%) → +USD 400,000 on the deal
Deviation
0%

🔗 Duplicate financing Critical

The invoice is hashed (SHA-256) into a document fingerprint and checked against the shared financing registry — the same method used in industry de-duplication networks. No document content is exposed.

fingerprint(MC-INV-2026-0847) =
computing…
Step 5 · Disposition

Trade Authenticity Score

Every signal is weighted into one defensible number, with the full reasoning attached for the audit trail.

0%
Fraud probability
Authenticity / 100
Active flags
Recommended action
HOLD TRADE — Manual Review
Do not release funds. Escalate to trade compliance; request original carrier B/L and independent re-survey before any disbursement.

Why this score

Weighted contribution of each signal to the fraud probability.
Audit trail