Case studies/Logistics and freight
A regional 3PL running 11 distribution centres, 16,000 proofs of delivery a month, and a billing close-out that used to slip 48 to 72 hours past the delivery date. We built a pipeline that ingests dock scans, fax queues, and driver phone photos into one stream, matches against the load tender, and posts into McLeod same-day.
| Load | Driver | Consignee | POD source | Match | Status |
|---|---|---|---|---|---|
| LD-2026-04918 | R. Alvarez | Heartland Grocers | Dock scan | 0.97 | posted |
| LD-2026-04919 | M. Okafor | Cedar Valley Foods | Driver photo | 0.94 | posted |
| LD-2026-04920 | T. Bergstrom | Northwind Retail | Signed BOL | 0.91 | pending |
| LD-2026-04921 | J. Rhee | Prairie Distribution | Dock scan | 0.96 | posted |
| LD-2026-04922 | D. McCarthy | Lakeside Supply Co. | Driver photo | 0.78 | review |
| LD-2026-04923 | S. Patel | Bluestem Foodservice |
| Dock scan |
| 0.95 |
| posted |
| LD-2026-04924 | K. Whitfield | Ironwood Logistics | Driver photo | 0.92 | pending |
| LD-2026-04925 | L. Petrov | Summit Grocers | Dock scan | 0.98 | posted |
At a glance
Eleven distribution centres, one billing close-out, and a driver population that does not read process documents. The pipeline had to meet them where they already work.
The engagement
The stack
ISO 27001 · ISO 9001 · DPA and NDA signed at kickoff.
Before, the billing desk
The billing team had a process. It worked. It also ran two days behind, because the PODs did not arrive in one place and nobody owned the last mile of the paperwork.
Drivers at the bigger DCs scanned at the dock. Drivers at the smaller DCs sent photos from their phones. Some shippers faxed back signed copies. The billing team stitched the channels together by eye, load number by load number.
Pre-build baseline: average 8 minutes per POD to locate the source and cross-reference the load tender.
Every POD had to be verified against the signed load tender for pieces, weight, and accessorial charges. The tender lived in McLeod; the POD lived in an email or a scan folder. Two windows, two logins, one spreadsheet.
Pre-build baseline: approximately 6% of PODs hit billing with a tender mismatch, caught only at close-out.
Because the POD arrived in four places and the tender lived somewhere else, the fastest any load could be invoiced was the next business day. On Friday deliveries, Monday morning. On a truck that delivered Thursday night, sometimes Tuesday.
Pre-build baseline: 48 to 72 hours from delivery to invoice raised, measured across the prior quarter.
What we built
The pipeline runs the same five stages we implement on every logistics engagement. The details below are the ones we wrote against this operator's dock behaviour, not a generic template.
Dock scanners drop to SFTP. Driver phone uploads route through a mobile form. Carrier email pulled on a 5-minute cadence. Fax queue OCR'd on receipt. All normalised to a single POD ID.
Document type tagged on ingest. Driver-photo PODs routed to LandingAI for handwriting. Dock scans routed to LlamaParse. Classification confidence below 0.88 holds the document for a dispatcher to tag.
Load number, pieces, weight, consignee signature, delivery timestamp, accessorial codes. LlamaParse primary, AWS Textract fallback when LlamaParse misses the SLA.
Pieces, weight, accessorials cross-checked against the signed load tender in McLeod. Signature present, timestamp inside the delivery window. Below 0.90 overall, the POD routes to the billing exception queue.
Matched PODs posted to McLeod via the load API with the source scan attached. The custom TMS receives the same event. Exceptions surface in a named queue with the flag in plain English.
After, the numbers billing signs off
Same drivers, same DCs, same shippers, same carrier relationships. The pipeline collapsed the paperwork stack and fed the billing team a clean record per load. The close-out moved because the data moved.
The billing team still signs off every invoice. Dispatchers still own the exception queue. The difference is that on a clean load, McLeod carries the POD and the matched tender by the time billing opens their queue. On a bad one, billing sees one flagged field, not a four-channel hunt.
From the billing desk
We kept our process. The pipeline just made the PODs show up where we needed them, on the day the truck delivered.
Billing operations leadRegional 3PL, Midwest hub
Handover
The engagement ends at a clean handover. The billing and dispatch teams run the pipeline; Hexaa stays on call for a fixed retention period, then steps back.
Related cases
Each links to a named client, a named document, and the system the clean data lands in. We publish only what the client signed off to publish.
28,000 customs packets a month reconciled inside a four-hour clearance window. LlamaParse on intake, CargoWise as the system of record.
→Logistics · 2024Mid-market broker · rate confirmations to the TMS6,200 rate confirmations a month matched to MercuryGate without the copy-paste round. Exceptions surface once, not across three tabs.
→Construction · 2025Commercial GC · submittal queue against the primeSubmittal log reconciliation against the prime contract spec. 1,200 submittals per project, Procore as the system of record.
→Free 30-minute call
You'll leave with a clear next step.
The dock scan says one set of pieces. The driver photo sometimes disagrees. The load tender is the authority. The pipeline compares all three at the field level and flags the mismatch before billing sees it.
