Loop.

A 4 a.m. shift starts at a Michael Kors distribution center in New Jersey. A trailer pulls in two hours late and 40 miles over its quoted route. By the time a human would normally notice, Loop already has - flagged the overcharge, paired it to the carrier contract, queued the dispute, and slipped the corrected number into the finance team's morning dashboard. Nobody at Loop touched a thing.

This is what Loop, the San Francisco logistics AI company, does on a Tuesday. Quietly. While the rest of the supply chain world argues about dashboards.

01The problem hiding in plain sight

For a sector that moves roughly $11 trillion worth of goods a year, freight runs on remarkably bad paperwork. Bills of lading get faxed. Invoices arrive as scanned PDFs of scanned PDFs. EDI feeds carry typos that nobody catches because nobody reads them. Carrier contracts hide on someone's hard drive, sometimes lawyered, sometimes scribbled.

Logistics, in other words, is the world's largest dark-data problem dressed up as a transportation problem. CFOs at Fortune 500 shippers have known this for years. They built audit teams. They bought BI tools. They hired consultants. The overcharges kept arriving anyway, in volumes too vast for any human queue to drain.

The conventional wisdom said you fix this by digitizing one piece at a time: cleaner TMS, better OCR, a freight-pay vendor here, a contract management tool there. Sensible. Also wrong. The mess is structural, not local. You can't fix it with a smarter form field.

02The founders' bet

Matt McKinney and Shaosu Liu met inside Uber Freight, where they spent enough time inside the engine room of modern trucking to understand exactly how leaky it was. McKinney - USC and Stanford engineering, a serial founder before he was 30 - had the operator's instinct. Liu had the systems instinct, the patience to build software that talks to other software when the other software was never really designed to talk.

In 2021 they started Loop with what looked, at the time, like an unfashionably narrow wedge: freight audit and pay. Audit the invoice. Pay the carrier. Catch the errors. The kind of company a VC pitch deck calls "back-office automation" and a CFO calls "the thing I lose sleep over."

The bet underneath the wedge was the interesting part. McKinney and Liu didn't think they were building a freight-pay company. They were building a data company that happened to enter through freight pay - because freight pay is where the messiest documents collide with real money, every single day, at scale. Solve that, and you've quietly built a training set the rest of logistics would kill for.

03The product, and the duck

At the center of Loop's platform sits DUX, a family of logistics-trained foundation models and agents. It is named like a duck. It behaves like a forensic accountant who never sleeps and never asks for a raise.

DUX ingests the documents that make a logistics team groan: rate confirmations, lumper receipts, accessorial charges, parcel manifests, EDI 210s, the handwritten correction stapled to page three. It standardizes them. It contextualizes them against carrier contracts. It compares them to what should have been billed. And then - this is the part nobody else has gotten right at scale - it acts. Loop's Exception Agent disputes the discrepancy, routes the payment, and updates the ledger. Touchless. The company reports more than 99% automation rates without errors.

Around DUX sits a wider platform: parcel visibility for e-commerce shippers, parcel contract optimization, Loop Intelligence for self-serve analytics, and a freight audit product with what the company claims is 99% modal coverage. It is, taken together, less a software product and more a layer - the intelligence layer that supply chain teams have been trying to assemble out of a dozen disconnected tools for a decade.

04The proof

A platform play only earns the name if real enterprises hand it real work. Loop's customer roster reads like a stress test of the consumer economy: Michael Kors handbags, Under Armour athletic wear, General Mills cereal, Kendra Scott jewelry, Great Dane trailers, GILLIG city buses, Loadsmart on the carrier side. Different industries, identical pain.

The numbers behind the logos are where the story sharpens.

Above: the parts of the story Loop's investors actually underwrite. Notice what is missing - a unicorn valuation. Loop hasn't disclosed one, and refreshingly, hasn't needed to.

The April 2026 round, led by Valor Equity Partners and its Valor Atreides AI Fund, brought in 8VC, Founders Fund, Index Ventures, J.P. Morgan Growth Equity Partners, and Tao Capital Partners. That last one matters more than the headline number. J.P. Morgan as a growth investor is not a financial bet. It is a distribution signal - a bank that touches half the enterprise treasuries in America wanting to be in the room when those treasuries get rewired around freight data.

05The mission, stated plainly

Loop describes what it is doing as "unlocking profit trapped in the operations of the physical economy." It is the kind of sentence that sounds like marketing right up until you spend an afternoon inside an enterprise freight team and realize it is, in fact, a fairly literal description of the job.

There is a quiet thesis underneath: that the next generation of enterprise value will not come from new software categories so much as from finally instrumenting the categories that were always there. Trucks moved before SaaS existed. Cereal got shipped before AI existed. The opportunity isn't to replace any of that. The opportunity is to read it - cleanly, continuously, and without humans drowning in PDFs to do it.

06Why it matters tomorrow

Forecasters love to talk about supply chain disruption. Pandemics, tariffs, canal blockages, port strikes - the playbook of the 2020s, told in headlines. The unspoken part is that companies don't lose money to disruption alone. They lose money to disruption plus the inability to know, in real time, what disruption is actually costing them. The first is unavoidable. The second is solvable. Loop is betting that solving the second one is a generational business.

There is competition, naturally. Legacy freight-pay vendors. Newer logistics SaaS upstarts. The big TMS incumbents who would like very much to add an AI sticker to their existing modules. None of them, so far, have built what Loop has built where Loop has built it: a logistics-trained model family with enough real freight data behind it to actually act, not just describe.

Back to the New Jersey distribution center. The trailer pulls in late. The overcharge gets caught before the coffee is brewed. Multiply that by a Fortune 500 portfolio, multiply that by a year, multiply that across every shipper who eventually plugs into Loop's platform, and you get a number large enough to explain why investors keep writing larger checks. You also get a slightly weirder result: a logistics back office that, for the first time, runs faster than the freight itself.

Loop didn't invent freight. It is, however, quietly inventing the part where the paperwork finally catches up.