Freight & Logistics

The freight claims you never file are worth more than the ones you argue about

Computer vision and AI agents for 3PLs, regional carriers, and mid-market shippers. We build the evidence and exception systems, and we are paid from the dollars they recover - not upfront.

Freight is one of the few operations where the value of an AI system is verified by somebody other than the vendor selling it. A recovered cargo claim is a dollar a carrier decided to pay after previously saying no. A billing cycle is a timestamp in your own TMS. That external verification is exactly why we will work in this vertical without an upfront fee.

Where the measurable money sits

1. Damage evidence at the dock door

Cargo claims are usually lost on evidence rather than on liability. The claim needs proof that freight was intact at one handoff and damaged at the next. In practice, the delivery receipt gets signed clean because the driver is waiting, photos live on somebody's phone with no timestamp or PRO number attached, and by the time the damage is noticed the chain of custody is unprovable.

Fixed cameras at dock doors solve the boring part: every pallet is photographed and timestamped against the shipment reference automatically, whether or not anyone remembers. That gives you an evidence packet for claims that would otherwise be denied, and - the larger number in most operations - it surfaces claim-eligible damage that currently goes unfiled entirely.

For scale, the 2025 Flock Freight Shipper Research Study, surveying 1,000 shippers, put the LTL damage rate at 1.24%, about one shipment in eighty, with an average LTL damage claim of $1,796. Carrier-level variance is enormous, with some reporting under 0.1%, so your own claim history is the only number that matters for scoping.

2. Days-to-invoice, not OCR accuracy

Freight document processing is a solved problem technically and an unsolved one operationally. General-purpose multimodal models now read a bill of lading well enough that nobody should pay a premium for extraction accuracy. What still costs real money is the cycle: documents that arrive out of order, invoices that sit because a POD is missing, rate confirmations that silently disagree with what got billed.

So we do not sell you OCR. We build the matching and exception workflow, and we attach the fee to days-to-invoice - a date already stamped in your TMS, which you control and we cannot influence except by actually shortening it.

3. Exceptions nobody has time to chase

Detention and demurrage that were never billed, accessorials that were absorbed, appointment misses that became a chargeback, shipments quietly aging past their claim filing window. These are individually small and collectively significant, and they are precisely the work that gets dropped when the team is busy. Agents that watch for them and assemble the paperwork convert dropped revenue into billed revenue.

Why we like this vertical for performance pricing

Most "AI saves you money" claims are unverifiable because the vendor and the client are both estimating the savings. Freight is different in one specific way: the carrier decides whether to pay the claim, and the payment shows up in your AR. We cannot inflate that number, you do not have to trust our arithmetic, and there is nothing to argue about at settlement time. When we say we will take a share of recovered dollars, that is a genuinely falsifiable promise.

What this is not

How an engagement starts

  1. Free audit. We look at your claim history, denial reasons, and billing cycle, and tell you which of the three levers above is worth the most in your operation - with a number attached.
  2. Written measurement agreement. The metric, the baseline, the attribution method, our percentage, and the cap. Signed before anything is built.
  3. Build and deploy at no upfront fee to you.
  4. You pay from what it recovers. If nothing moves, you owe no performance fee.

Full mechanics, including how the baseline and the cap work, are on our pricing page.

Sources

Every figure on this page is linked to its origin so you can check it. Where a number is vendor-published marketing rather than independent research, we say so on the page.

  1. Flock Freight, 2025 Shipper Research Study (survey of 1,000 shippers) - source of the 1.24% LTL damage rate and $1,796 average damage claim.
  2. Warp, LTL Damage Rates: Why Fewer Touches Means Fewer Claims - analysis of damage rates against the multi-touch terminal model, and carrier-level variance.

Founding client terms

We are early, we want proof, and we are pricing accordingly

Book the free audit We run three performance-based engagements at a time

Frequently Asked Questions

Why do LTL freight damage claims get denied?

Most denials come down to evidence, not liability. A cargo claim requires proof that goods were intact at one handoff and damaged at the next. When the delivery receipt was signed clean, when photos have no timestamp or shipment reference, or when nobody documented the freight before it left the dock, the carrier has a defensible reason to deny. Fixing evidence capture at the handoff is usually a bigger lever than arguing claims harder afterwards.

How does computer vision help with freight claims?

Fixed cameras at dock doors photograph and timestamp freight condition automatically at each handoff, tied to the shipment reference, without anyone remembering to do it. That produces a complete evidence packet for claims that would otherwise be denied, and it surfaces claim-eligible exceptions that currently go unfiled because nobody noticed in time.

How common is freight damage in LTL shipping?

The 2025 Flock Freight Shipper Research Study, a survey of 1,000 shippers, reported an LTL damage rate of 1.24% - roughly one shipment in eighty - with an average LTL damage claim of $1,796. Carrier variance is wide, with some reporting under 0.1%, and damage scales with the number of handoffs, so cross-dock and multi-stop lanes carry more exposure.

Is AI document extraction still worth paying for?

Not on accuracy alone. General-purpose multimodal models read freight documents well enough that paying a premium for OCR no longer makes sense. The durable value is the workflow around it - matching documents to loads, catching rate-confirmation mismatches, closing the billing cycle faster - which is why the honest metric is days-to-invoice, a timestamp already in your TMS.

How would a performance-based fee actually be measured here?

The counterparty verifies it. A recovered claim is a dollar amount a carrier chose to pay after previously denying it or never receiving it, and it lands in your AR. Days-to-invoice is a timestamp in your own system. Neither number is estimated by the vendor being paid, which is what makes the arrangement auditable rather than a matter of trust.

Do you need access to our TMS to start?

No. These systems usually start as a parallel evidence layer with read-only access or no system access at all, which is why they can be deployed quickly and without an IT project. Integration comes later, once the value is already measurable.

Find out what your denied claims are worth

Free audit of your claim history and billing cycle. If the number is too small to bother with, we will say so.

Schedule the Free Audit

Other operations we work in