A shipping container is a wonderfully honest object. It has hard edges, fixed volume, a weight limit, and no sympathy for pitch decks. Benjamin Martindale has built his career around what happens before one is sealed and after another is opened. His latest company, Donkey, begins with an almost mischievously simple ambition: make ordering a container feel less like commissioning a minor naval expedition and more like filling a cart.
The idea carries a biography inside it. Martindale grew up in his family's industrial supply business. The products were physical, the buyers practical, and the distance between a quoted price and a delivered item could contain a small civilization of brokers, duties, freight bills, inspections, and unwelcome surprises. He later studied finance at Rotman Commerce at the University of Toronto, leaving one credit short. It is the sort of detail that feels less like a polished credential than an honest corner left showing.
He learned Mandarin and moved closer to the machinery. He advised Chinese heavy-equipment companies Sany and Sinoboom on expansion into Western markets. He appeared in factory-floor interview segments and a 13-minute profile on Mango TV. Before that chapter became Donkey, he ran a Toronto marketing company that closed a reported $4 million for construction clients in six months. The through-line is not technology. It is selling complicated things to people who need the complication removed.
Thirty-one yeses in Guangzhou
Prelude OS was the first clean expression of that instinct. The company built trade-intelligence software for Asian manufacturers: read American customs filings, identify the US companies already buying a given product, find the relevant decision-makers, and help a factory write credible English outreach. The software translated more than language. It translated a shipping record into a possible customer.
Martindale did not test the proposition from a Toronto desk. On the ground in Guangzhou, he cold-acquired 31 paying Chinese factory customers in 39 days. Prelude became cash-flow positive. Those numbers are compact enough to fit in a founder bio, but they imply a lot of doors, messages, demonstrations, objections, and cups of tea. Factory owners are not buying the romance of software. They are buying a credible route to an order.
Prelude looked east across the Pacific, helping factories find American buyers. Donkey faces west, helping American buyers reach factories. The two businesses sit on opposite sides of the same container. One helps the maker locate demand. The other helps the buyer locate supply and then get it home without becoming an amateur customs broker.
“We've lived both ends of the container.”Benjamin Martindale and his co-founder, on launching Donkey
A cart with ocean freight
Donkey's opening argument is economical: everything gets cheaper by the container. Large companies buy the box. Smaller businesses often buy from someone who bought it, absorbing the markups along the way. Martindale's answer is to let a buyer upload a list of products and see a single landed price, with tariffs and delivery included. The software packs the order against a container's volume and payload. If the load spills over, another box appears in the plan.
This is where the company becomes more interesting than a sourcing directory. The importer buys from Donkey. Donkey takes title, coordinates the factories, puts inspectors on the floor before payment, handles the freight and duty, and delivers to the buyer's dock. When the buyer signs, the number holds for the stated period. If a cost underneath that number moves, Donkey owns the headache.
The division of labor inside the product is unusually clear. AI agents read public customs records, separate likely factories from shells and forwarders, and help pull quotes. Duties are then applied line by line. The final pricing math is deterministic. In a year when almost any arithmetic can be decorated with artificial intelligence, choosing to say where AI stops is a small act of good manners.
That separation suits the physical world. A language model can be usefully uncertain while reading an abbreviated customs description. A builder signing for cabinets, flooring, doors, and bathroom fixtures has less appetite for creative ambiguity in the duty calculation. Martindale's product places probabilistic tools at the fuzzy beginning of the job and fixed arithmetic near the invoice.
The first public box
Donkey's early proof point is a home builder in Oakland. The order placed 15 SKUs into one container and produced a reported saving of 38 percent against buying the items piecemeal from distributors. The company's example price was $51,318.92 delivered, compared with published big-box lists north of $73,000 for the same or equivalent specifications.
One order is a case, not a law. It is still a useful case because it shows the customer Donkey is beginning with: people who build. Construction procurement is full of repetitive products, significant markups, fragmented buying, and schedules that punish surprises. A cabinet door arriving late does not care that the sourcing dashboard was elegant.
The 15-SKU detail also matters. A small importer may know that a full container offers better economics and still lack enough demand for a container of any single item. Mixed loads change the calculation. Six cabinet sets can travel with doors, levers, pulls, flooring, vanities, and toilets. The buyer aggregates a project rather than a product. Software does the awkward packing work, checking cubic meters and weight as quantities move. What appears on screen as a responsive shopping cart is, underneath, a geometry problem with tariffs.
The approach shifts the comparison away from a factory's attractive unit price. A low figure at the factory gate can become considerably less charming after ocean freight, customs entry, duties, inland trucking, inspection, and financing arrive. Donkey puts those costs beside each line and then offers the whole load as one delivered-duty-paid number. The buyer can compare a price at the dock with another price at the dock. Martindale's finance education has found a practical home here: the important number is not the first one in the chain, but the one that survives all the way to the end.
Martindale's background gives him a natural fluency with this customer. Industrial supply was family terrain. Construction companies were clients of his marketing business. Chinese manufacturers were customers and advisory partners. Donkey is the point where those relationships meet, now with a seven-person team split between San Francisco and China and a co-founder building the data machinery underneath the trade.
“Everything gets cheaper by the container.”Donkey's opening proposition
The software is responsibility
There is a popular version of an AI company that ends at the recommendation. It finds a supplier, summarizes a record, or drafts a message, then leaves the difficult nouns - money, goods, damage, delay - to somebody else. Donkey has chosen the more awkward route. Its service is only simple to the buyer because the company absorbs coordination behind the screen.
The choice explains why Martindale calls it a trading company rather than merely a marketplace. A marketplace introduces. A trading company stands in the transaction. It can improve the price as volume grows, but it also has to know which factory will deliver, which inspection will catch a bad batch, and which promise can survive a week of shifting freight rates.
Taking title changes the incentives. A lead-generation platform can celebrate when two parties meet. Donkey has to care what goes into the carton. The company says its own inspectors are on factory floors before a factory gets paid, and the photographs on its site carry the founder's credit. That detail fits Martindale's path. He has not approached trade as a purely digital abstraction. His biography repeatedly returns to factories, whether he is selling software in Guangzhou, advising equipment makers, or documenting packed pallets. The physical evidence is part of the interface.
Credit adds another layer. Qualified buyers can receive net-45 terms backed by credit insurance. For a small importer, paying after receiving goods can matter as much as shaving dollars from a unit price. For Donkey, it means the product includes money moving at different speeds: factories expect payment, ships take time, and customers preserve working capital. The neat quote on the screen therefore compresses not only logistics but timing. Someone has to finance the interval between an inspected pallet and a paid invoice.
His earlier work was about revealing relationships hidden in customs data. Donkey adds a second move: acting on them. Public filings can show that a distributor bought a product, from whom, and through which lane. That knowledge may expose a margin. It does not, by itself, arrange a quality check in Guangdong or put the right pallet on a truck in Oakland. Information becomes valuable when someone carries it across the gap.
Martindale has spent years crossing that gap in both directions, between Canada, China, and now San Francisco; between manufacturers and buyers; between software's clean screen and a factory floor's useful mess. The aspiration behind Donkey is not to make international trade look glamorous. It is to make it feel ordinary.
That may be the better ambition. The container transformed commerce because it standardized the box. Martindale is trying to standardize the experience around it: bring a list, receive a number, know who is responsible. The cart is a playful metaphor. The accountability is the product.