Every importer has been there. You're buying widgets from a supplier in China, paying what seems like a fair price, and somewhere in the back of your mind sits a nagging question: Am I talking to the factory—or am I just enriching a middleman?
Donkey, a startup that emerged from Y Combinator, is wagering that question keeps enough people up at night to support an entire business model. The company's proposition borders on the audacious: snap a photo of any product you import regularly, and within 72 hours Donkey will tell you which Chinese factory actually makes it, along with a delivered-to-your-dock price that cuts out the intermediaries.
It's a pitch that speaks to a fundamental frustration in cross-border trade. The opacity. The markup uncertainty. The sense that someone, somewhere in the supply chain, is collecting rent you can't quite see.
Whether Donkey can deliver at scale remains an open question—one the founders are essentially betting their company on.
Customs Data as a Decoder Ring
The mechanics are less mystical than they sound. Donkey's platform doesn't rely on supplier directories or self-reported factory certifications. Instead, it mines US customs records—millions of import filings that list shippers, consignees, product descriptions, and cargo values. Those documents, public but sprawling, contain patterns that can separate actual manufacturers from trading companies and logistics middlemen.
Minghao Tan, Donkey's co-founder and CTO, built the software that performs that parsing. It's a data-engineering challenge that requires filtering noise, disambiguating company names, and cross-referencing shipment histories to identify who's sending goods versus who's just moving paper. The company calls itself "AI-native," a bit of startup jargon that in this case points to machine learning models trained to recognize factory signatures in customs filings.
Once a buyer uploads a product photo, Donkey's system attempts to match it against known shipment records, pinpoint the originating manufacturer, and generate a quote. The company then takes on the full import process: negotiating with the factory, arranging inspection, coordinating freight, clearing customs, and delivering the goods. Donkey acts as the Importer of Record—meaning it holds title to the cargo until it reaches the buyer's facility—and quotes a single DDP (Delivered Duty Paid) price.
That structure collapses what's typically a multi-party negotiation into one transaction. No separate freight forwarder invoice. No customs broker bill. No ambiguity about who's responsible if something goes sideways at the border.
The company says it calculates tariffs on what it describes as "the honest factory invoice" rather than inflated trading-company numbers—a detail that matters more as duty rates climb. Donkey also deploys its own inspectors across seven provinces in China, withholding payment to factories until quality checks clear. Live tracking and inspection photos are baked into the platform, an effort to give buyers visibility they don't always get through traditional channels.
A Founder Who Knows the Customs Playbook
Benjamin Martindale, Donkey's other co-founder, has been working the customs-data angle for a while. Before Donkey, he built Prelude OS, a trade-intelligence tool aimed at Asian manufacturers. That platform uses US import records to show factories which American companies are already buying their type of product—essentially a lead-generation engine powered by public filings. According to Prelude's materials, it has profiled tens of billions of dollars in import volume and scored thousands of US buyers algorithmically, though the exact figures cited in earlier materials may reflect an earlier snapshot.
Martindale speaks Mandarin and has done consulting work for Chinese heavy-equipment makers. It's the kind of background that gives him fluency in both the language and the unspoken rules of Chinese manufacturing—where relationships matter, where trust is earned slowly, and where a Western buyer's naïveté can be expensive.
Tan handles the technical plumbing. Y Combinator's description of the company notes it runs "with a handful" of people, leaning heavily on automation. The YC directory lists seven employees, though LinkedIn indicates a broader range of 11–50—the kind of ambiguity typical of early-stage startups, where headcount fluctuates and roles blur.
Donkey was founded recently and is working through the standard startup gauntlet: proving the model, refining the software, scaling operations without breaking them.
Timing and Tariff Turbulence
The company's launch coincides with a particularly messy chapter in US-China trade policy. Tariff structures have shifted repeatedly in recent years, with Section 301 levies on Chinese goods still in place and additional surcharges cycling through statutory expirations and renewals. A temporary measure under Section 122 of the Trade Act is scheduled to expire on July 24, 2026, though the broader tariff architecture remains.
US imports from China totaled $308.4 billion in 2025, down 29.7% from the prior year but still representing a massive flow of goods. For importers navigating that landscape, every percentage point of markup—or every dollar of duty paid on an inflated invoice—adds up quickly.
Traditional B2B marketplaces like Alibaba.com and Made-in-China.com offer buyer protections and, in some cases, AI-powered sourcing tools. But those platforms still depend largely on supplier self-reporting. Distinguishing a real factory from a trading company often requires manual vetting—factory visits, business-license checks, reference calls. Sourcing agents and middlemen typically charge commissions between 5% and 10%, or apply markups of 15% to 30% or more, depending on the product and relationship.
Donkey's model attempts to bypass that layer entirely by embedding its margin into a single delivered price and assuming the importer-of-record role directly. Under Incoterms 2020 (the international commercial terms that govern trade), DDP is the most seller-friendly arrangement—placing maximum responsibility on the party shipping the goods. Donkey is betting that convenience is worth the premium.
The Hard Part: Making It Work at Scale
Donkey's website is live at donkey.trade, with detailed walkthroughs and FAQ sections. The company has not disclosed funding amounts beyond its Y Combinator affiliation, and no major customer wins or partnerships have been announced publicly in available materials from the relevant period.
The pitch itself is clean. Upload a photo. Get a factory-direct quote in three days. Let the platform handle everything else. For importers accustomed to supplier ambiguity—wondering if they're overpaying, questioning whether their "factory contact" is really just a trading-company rep—the appeal is obvious.
But the hard part isn't the pitch. It's execution. Can Donkey's data engine accurately identify manufacturers from a product image and a haystack of customs filings? Can it scale inspection operations without quality slipping? Can it negotiate competitive pricing at volume, especially for buyers outside its initial target categories—housewares, furniture, tools, lighting, automotive parts?
And perhaps the trickiest question: can it build trust with American importers who've been burned before, in an industry where relationships and reputation move slowly?
The company is making a direct bet on transparency in a corner of global trade that has resisted it for decades. Whether that bet pays off will depend less on the elegance of the software and more on whether Donkey can execute consistently, shipment after shipment, as volume grows. The model is intriguing. The proof will be in the containers that clear customs.
