Basically no one is using AI agents for their DoorDash orders

As AI tools and agents have continued their march across all aspects of digital life, many companies are trying to use LLMs to improve the experience of purchasing goods and services on the Internet. Thus far, though, a vanishingly small number of people seem to be taking advantage of those capabilities, at least when it comes to using the popular food delivery app DoorDash with an outside AI agent.

随着人工智能工具和智能体在数字生活的各个方面不断推进，许多公司正试图利用大语言模型（LLM）来改善在互联网上购买商品和服务的体验。然而到目前为止，似乎只有极少数人利用了这些功能，至少在使用流行的外卖应用 DoorDash 配合外部人工智能智能体时是这样。

In a third-quarter earnings call Thursday, DoorDash co-founder and CEO Tony Xu responded to an analyst question by noting that “AI agents are generating several hundred orders [on DoorDash] across the time that they’ve been up and running.” In a marked understatement, Xu called this a “very low use case” for a company that handled just shy of 1 billion orders total in the third quarter of 2026 alone.

在周四的第三季度财报电话会议上，DoorDash 联合创始人兼首席执行官托尼·徐（Tony Xu）在回答分析师提问时指出：“人工智能智能体在上线运行期间，在 DoorDash 上仅产生了数百个订单。”徐以一种明显的轻描淡写称，对于一家仅在 2026 年第三季度就处理了近 10 亿个订单的公司来说，这是一个“非常低的使用案例”。

Speaking generally, Xu said he sees the potential for AI agents to be genuinely helpful for “long consideration purchases” and major decisions where agents would be “doing work on your behalf proactively in the background.” However, Xu added, “when it comes to a task that’s as simple as ordering food, we tend to see that come directly to DoorDash.”

从总体上讲，徐表示他看到了人工智能智能体在“需要长期考虑的购买”和重大决策方面的真正潜力，在这些情况下，智能体可以“在后台主动代表你工作”。然而，徐补充说：“当涉及到像点餐这样简单的任务时，我们往往看到用户直接通过 DoorDash 进行操作。”

To be fair to the AI boosters, DoorDash only rolled out its Model Context Protocol (MCP) connector less than a month ago. The connector provides a standardized, open-standard way for third-party AI tools to easily order through DoorDash’s platform. Without that kind of linkup—which is still only available to a small number of “early beta partners” and mainly used to streamline corporate lunch orders—AI agents have to go through a bit more effort to access DoorDash through a web interface, for instance.

公平地说，DoorDash 在不到一个月前才推出了其模型上下文协议（MCP）连接器。该连接器为第三方人工智能工具提供了一种标准化的开放标准方式，以便通过 DoorDash 的平台轻松下单。如果没有这种连接——目前仅提供给少数“早期测试合作伙伴”，且主要用于简化公司午餐订单——人工智能智能体就必须付出更多努力，例如通过网页界面访问 DoorDash。

Xu’s discussion of “third-party agent” orders also doesn’t include DoorDash’s own internal AI tools. Those include an “Ask DoorDash” feature the company rolled out this summer that can, for instance, create an online shopping cart from a photograph of a recipe or suggest an order based on a vague description like “dinner for a family of four.”

徐关于“第三方智能体”订单的讨论也不包括 DoorDash 内部的人工智能工具。其中包括公司今年夏天推出的“Ask DoorDash”功能，例如，该功能可以根据食谱照片创建在线购物车，或根据“四口之家的晚餐”这样模糊的描述来建议订单。

DoorDash hasn’t shared detailed statistics on that kind of usage, but the company said last month that Ask DoorDash baskets get built “five times faster” than those built without, and Ask DoorDash orders “average nearly 50% higher basket value and about 60% more unique items than the same consumers’ traditional orders.” The company also said nearly half of Ask DoorDash orders end up at “local spots the consumer has never tried before,” helping with discovery of new area restaurants.

DoorDash 尚未分享此类用法的详细统计数据，但该公司上个月表示，使用 Ask DoorDash 构建购物车的速度比不使用时“快五倍”，并且 Ask DoorDash 订单的“平均客单价高出近 50%，独特商品数量比同一消费者的传统订单多出约 60%”。该公司还表示，近一半的 Ask DoorDash 订单最终流向了“消费者从未尝试过的当地餐馆”，这有助于发现新的区域性餐厅。

That’s all to say that DoorDash’s current experience with agentic ordering may be a bit like Pizza Hut’s first tentative online ordering experiments in 1994. Back then, it would have been hard to picture today’s world, where more than half of Pizza Hut orders are placed online. Maybe ordering food and other goods using AI agents will be just as commonplace years from now, and we’ll look back at these early statistics with bemusement.

总而言之，DoorDash 目前在智能体下单方面的体验可能有点像必胜客 1994 年首次尝试在线订餐。当时，人们很难想象今天的世界，即超过一半的必胜客订单是在线完成的。也许几年后，使用人工智能智能体订购食品和其他商品将变得同样普遍，而我们将带着困惑回顾这些早期的统计数据。

For now, though, Xu at least seems relatively cool on the early numbers. “My take on all things technology is you never want to fight gravity,” he said regarding AI agents. “But at the same time, you don’t want to be delusional.”

不过就目前而言，徐对这些早期数据似乎相对冷静。他在谈到人工智能智能体时说：“我对所有技术的看法是，你永远不要试图对抗重力。但与此同时，你也不想自欺欺人。”