Industry
Every load generates a paper trail — rate confirmation, BOL, POD, carrier invoice, and a running thread of status emails — and right now your people key most of it by hand. We build automations that read those documents and draft the entry, the match, and the billing packet; your team reviews and approves before anything touches your TMS or your accounting system. The architecture is proven in live manufacturing and packaging operations — shipment tracking, invoice reconciliation, quote processing — and it translates directly to freight.
The grind
Every carrier invoice should be checked line by line against the rate confirmation, but nobody has time — so detention charges, lumper fees, and fuel surcharge errors either slip through or get paid because auditing a $75 discrepancy costs more than eating it.
You can't bill the customer until the signed POD shows up, and it rarely shows up on its own. Someone on your team spends afternoons emailing carriers for paperwork on loads that delivered a week ago, while your receivables age for no operational reason.
Check calls, carrier status emails, and "where's my load?" messages from customers all land in inboxes, and your ops people spend the day copying the same status from one place to another — into the TMS, into a reply, into a spreadsheet.
Substantiating a detention claim means digging through email threads for arrival and departure times, appointment confirmations, and who said what. An hour of reconstruction to recover a $150 accessorial is a losing trade — so the charge gets paid, and the carrier learns your invoices don't get audited.
The split
We automate
Carrier invoice audit — line items extracted and matched against the rate confirmation and POD, with unagreed accessorials flagged for review
POD chasing and billing packets — follow-ups drafted, documents filed against the load, customer invoices staged for your biller's approval
Track-and-trace email triage — carrier status emails read and turned into draft TMS updates and draft customer replies
Rate confirmation entry — lane, rate, dates, and terms pulled from emailed rate cons into a draft load record for dispatch to confirm
Detention dispute files — arrival and departure times, appointment records, and correspondence assembled into a claim your team decides to send — or not
We protect
Carrier selection and rate negotiation — which carrier gets the load is judgment built on lanes, history, and trust — no algorithm makes that call
Pay, short-pay, or dispute decisions — every disputed accessorial touches a carrier relationship; a person makes that call, every time
Customer escalations and exception loads — when a load goes sideways, your people work the problem — with the full paper trail already assembled for them
Final approval before anything posts — no payment, TMS update, or customer invoice leaves the approval queue without human sign-off — by design, not as a phase
How it runs
AI drafts. A person reviews and approves. Nothing touches a live system without a human gate — that's the architecture, not a phase.
The full build architecture →A worked example
Your clerk's existing audit doesn't stop the day this starts — she keeps spot-checking invoices the old way until the queue has caught enough real discrepancies to earn her trust.
and your ops people keep working live loads by hand alongside it, so the judgment a freight desk runs on stays in daily practice.
We won't claim live freight deployments — we don't have them, and we'd rather say that plainly. What we have is a system running live at a North American manufacturing and packaging company that reads unstructured documents, matches them deterministically against expected values, and routes everything through human approval: vendor quotes, shipment tracking, invoice reconciliation, commission calculations. A carrier invoice against a rate confirmation is the same problem as a vendor invoice against a purchase order — the documents change, the architecture doesn't.
Plenty. EDI covers the carriers who invest in it — everything else arrives as email and PDF: rate cons from smaller carriers, emailed status updates, scanned PODs, invoices with hand-typed accessorials. Our automations read that unstructured layer and draft entries into the TMS you already run, for your team to approve. We integrate with your systems on your servers; we don't replace them.
We sit with your team, map how the work actually happens, and draw the split: what's worth automating, and what automation should never touch. You keep the map either way.
Map your workflows