Skip to content

// Selected Logistics Projects

WHAT WE'VE
ACTUALLY BUILT

Real systems built for real operations. Every build was designed around the client's actual trade documents, installed on their infrastructure, and handed over completely. Client details are anonymized by agreement — the problems and results are real.

10h+

staff hours recovered
per week, per client

−78%

compliance errors
in first quarter

97%

extraction accuracy
vs. manual baseline

<2wk

average build time
scoping to handover

01
Freight Brokerage Ontario, Canada Invoice Audit + Carrier Performance

FREIGHT BROKERAGE — INVOICE AUDIT SYSTEM

A mid-size freight brokerage processing 400+ carrier invoices per week was spending 12 hours of staff time per week on manual invoice verification. Discrepancies were being caught inconsistently — and often only after payment had already run.

The problem

Carrier invoices arrived in multiple PDF formats. Staff were manually comparing them to rate confirmations in a separate TMS. Accessorial charges were being accepted without verification because there was no practical way to check them all at volume. Duplicate billing was going undetected.

What we built

A private Invoice Auditor fine-tuned on their specific carrier invoice formats — covering 14 carriers across their network. The system was integrated into their existing TMS workflow via a simple file drop. Each invoice is processed in under 3 seconds, with a structured discrepancy report generated automatically.

Results after 30 days

−94%

billing discrepancies
reaching payment

12h

staff hours recovered
per week

$38k

in overbilling recovered
in first 60 days

<3s

per invoice
processing time

Build specs

Invoice Auditor + Carrier Performance module. Installed on their existing Windows server. Full source code handover. Build time: 8 business days from scoping to deployment.

02
Import/Export Rome, Italy Trade Document Processing

MEDITERRANEAN IMPORTER — TRADE DOCUMENT ENGINE

An Italian import operation handling mixed-commodity shipments from Asia and North Africa was managing 200+ trade document packages per month. Document inconsistencies causing customs holds were costing them an average of €4,200 per incident in delay costs and fines.

The problem

Commercial invoices, packing lists, and certificates of origin were being prepared by multiple parties — the exporter, the freight forwarder, and the local agent. Discrepancies in HS codes, declared values, and quantity descriptions were only found during customs clearance, resulting in holds that delayed delivery by 5–10 days.

What we built

A Trade Doc Engine fine-tuned on their specific commodity types and EU customs entry formats, with multi-language support for supplier documents arriving in English, Italian, and Mandarin. Processes the full document set in under 60 seconds and flags discrepancies before the package is submitted to customs.

Results after 60 days

−91%

customs holds
in first two months

8h

staff hours saved
per week

€47k

estimated savings
in delay cost avoidance

3

languages handled
without manual translation

Build specs

Trade Doc Engine + Compliance Guard modules. Deployed on their on-premise server in Rome. Multi-language (EN/IT/ZH). Build time: 11 business days. Full source code handover.

03
3PL / Supply Chain Toronto, Canada Carrier Performance + Rate Intelligence

CANADIAN 3PL — CARRIER INTELLIGENCE PLATFORM

A Toronto-based third-party logistics provider managing 30+ carrier relationships was doing quarterly carrier reviews manually — pulling exception data from their TMS, building spreadsheets, and comparing rates in Excel. The process took 3 days per quarter and was consistently missing high-performing and underperforming patterns between quarterly reviews.

The problem

No systematic visibility into carrier performance between quarterly reviews. Rate negotiation conversations were happening without lane-level benchmark data. Claim rate patterns were being identified too late — usually only after a shipper complaint prompted a manual audit.

What we built

A Carrier Tracker + Freight Rate AI system that ingests POD documents, exception reports, and carrier rate quotes from their existing TMS export. Generates automated weekly carrier performance reports, flags emerging performance issues before they affect shippers, and produces all-in rate comparisons across contracted lanes.

Results after 90 days

18%

improvement in
on-time delivery rate

3d

quarterly review time
reduced to 45 minutes

−62%

claim rate for
flagged carriers

$21k

annual rate savings
from renegotiation data

Build specs

Carrier Tracker + Freight Rate AI modules. Integrated with their existing TMS via file export. Deployed on their on-premise server. Build time: 10 business days. Full source code handover.

04
Air Cargo / Customs Broker Mississauga, Canada Entry Validation + HS Code Intelligence
Air cargo customs operation

// 04 — Air Cargo Customs Broker

CFIA & CBSA ENTRY VALIDATOR FOR A LICENSED CUSTOMS BROKER

A Toronto-based customs brokerage clearing ~140 air cargo entries per day was running into recurring CBSA AMPS penalties — averaging $11–14k per quarter — on entries that should have been catchable at intake. Tariff classifications were being pulled from a six-year-old internal lookup; valuation declarations weren't reconciled against the commercial invoice line totals.

The problem

Three licensed customs specialists were eyeballing a 6-document entry package per shipment — commercial invoice, packing list, AWB, certificate of origin, broker worksheet, and the importer's COO record. HS classifications were being reused from prior entries without re-checking the current CBSA chapter notes, and CFIA-regulated commodities were being misflagged on busy days. Errors only surfaced after the entry was filed.

What we built

A CBSA & CFIA Entry Validator running on their on-prem broker workstation, fine-tuned on their last 18 months of cleared entries. Ingests the full document package, verifies HS classification against the current CBSA Customs Tariff (Schedule I), reconciles declared values line-by-line, flags CFIA / OGD requirements before transmission, and produces a one-page broker review sheet. Integrated with their existing CADEX submission workflow via a watched folder.

Results after 90 days

−87%

AMPS-eligible
entry errors

$42k

penalty exposure
avoided (annualized)

2.3min

per entry review
down from 18 min

14h

specialist hours
recovered per week

Build specs

Entry Validator + HS Code Intelligence modules. On-prem deployment, no CBSA/CFIA data leaving the workstation. Tariff chapter updates pulled from public CBSA feeds on a daily cron. Build time: 13 business days. Full source code & chapter-update playbook handed over.

// Across every logistics build

WHAT EVERY
BUILD SHARES

Trained on your real documents

Every system is fine-tuned on the client's actual invoices, AWBs, BOLs, rate confirmations, and customs paperwork — never on synthetic or off-the-shelf training data.

Deployed on your infrastructure

On-prem or your VPC. No carrier rates, no consignee data, no commercial values ever leave the perimeter — not even for inference.

Handed over completely

Full source, model weights, the document-set we trained on, and an internal-team training session. Your IT can run and update it without us.

Bought once, owned outright

One-time build fee. No per-shipment, per-entry, per-seat, or per-month fee. The system is a capex line on your books, not an opex subscription.

// Your operation next

WANT TO SEE WHAT
WE CAN DO WITH YOUR DOCS?

Send us one real document. We'll show you exactly what the AI can surface from it within 24 hours — written report, no obligation, no pitch on the other end.

Request Your Free 24h Assessment →