When Benjamin Martindale tells the story of his startup, he skips the usual founder mythology and goes straight to containers. His company, Donkey, wants to make importing from China feel less like navigating a byzantine trading system and more like filling an online cart — complete with a price that includes tariffs, freight, and every hidden fee that typically surfaces weeks after a buyer thinks they've struck a deal.
The San Francisco–based company emerged from Y Combinator's Summer 2026 cohort with a straightforward pitch: upload a buying list, get a locked-in Delivered Duty Paid price within 72 hours, then let Donkey handle everything from factory vetting to customs clearance. Unlike Alibaba.com or Made-in-China.com, which connect buyers to suppliers without touching the merchandise, Donkey takes title to the goods. It becomes the seller, not the middleman.
That positions the startup as a trading principal in a market where most digital platforms have deliberately avoided that role. Taking ownership means taking on risk: quality disputes, shipping delays, and the working capital needed to float inventory between factory and buyer. Martindale, who grew up in a Toronto importing family and later advised Chinese heavy machinery firms like Sany and Zoomlion, appears to be betting that transparency and speed will outweigh those complications.
The platform pulls its data from public customs filings, using what the company describes as AI agents to parse shipment records and separate legitimate manufacturers from shell companies and freight forwarders. Once a buyer submits a request — via Excel spreadsheet, PDF, or even a photo of a handwritten list — Donkey's system identifies factories, pulls quotes, calculates duties line by line, and adds freight costs. The result: a single number that locks for 48 hours after the buyer signs.
"The pricing math is deterministic, not AI, so the number you see is the number you pay," Martindale said in the company's YC announcement, posted around August 20.
Whether that promise holds at scale remains an open question. The company's terms of service, updated on August 18, clarify that quotes are estimates and that binding purchases occur only when both sides confirm a written order. That's standard legal hedging, but it introduces a gap between the platform's frictionless pitch and the contractual reality.
Donkey says it inspects goods on factory floors before manufacturers get paid, a quality-control step that traditional trading houses have long performed but that digital marketplaces typically leave to buyers. Payment terms are structured around that checkpoint: qualified buyers get Net-45 credit insurance, with the balance due upon acceptance at the receiving dock. The company's LinkedIn page, which listed six employee profiles in early September, suggests a small team of five managing what is essentially a vertically integrated supply chain operation.
Martindale studied finance at the University of Toronto's Rotman School of Management and previously ran a Toronto marketing firm that reportedly closed $4 million in construction deals over six months. His cofounder, Minghao Tan — referred to as James on the company's homepage — focuses on the data infrastructure, building the agent systems that parse customs filings and map industrial supply chains. Tan's background includes work at InstaLILY, though details on that experience are sparse.
The startup is starting with construction buyers, a sector where price volatility and long lead times have historically made overseas sourcing a headache. A marketing example on Donkey's website describes an Oakland home builder who saved 38 percent on 15 items by consolidating cabinets, vinyl plank flooring, and iron stair railings into a single order. No other customers are named publicly.
The timing of Donkey's launch is worth noting. Section 301 tariffs on Chinese imports remain in place following a four-year review by the U.S. Trade Representative that wrapped in May 2024. Meanwhile, Customs and Border Protection revised its enforcement methodology under the Uyghur Forced Labor Prevention Act in January 2026, shifting from aggregated entries to shipment-level tracking. That change has added layers of compliance scrutiny for importers, creating both friction and opportunity for platforms that promise to navigate the complexity.
Freight forwarders like Flexport offer logistics coordination and customs brokerage but stop short of acting as trading principals. Landed-cost calculators such as Zonos guarantee duty and tax calculations without taking title to goods. Traditional trading houses like Li & Fung have the infrastructure to coordinate suppliers but have not built AI-native quotation systems pulling from real-time customs data. Enterprise procurement platforms like Keelvar and Arkestro use AI for sourcing optimization but position themselves as software vendors, not merchants.
Donkey's model combines elements of all these players while sidestepping their limitations — or at least that's the theory. Whether small and mid-sized buyers will trust a startup with their purchasing dollars, and whether Donkey can manage the operational complexity of being a trading principal at scale, are questions that will likely take longer than 48 hours to answer.
The company updated its LinkedIn status in early September to announce it was live and ready for buyers to submit lists. Y Combinator held its Summer 2026 Demo Day on September 10, with Harshita Arora listed as Donkey's primary partner at the accelerator. No investors or funding rounds beyond YC have been disclosed, and the company's terms specify that its service is intended for businesses, not individual consumers.
For now, Donkey remains a bet on a simple premise: that importers are tired of surprise costs and are willing to let someone else own the complexity in exchange for a number they can actually trust. Whether that number pencils out in practice is the story that comes next.
