DoorDash CEO Tony Xu said on the company’s third-quarter earnings call that DoorDash has seen ‘several hundred’ orders from third-party AI agents such as Muse, Instinct and Fo, according to TechCrunch. The figure is small next to the 3 million deliveries the company said it completes on an average day in its 2025 report.
What did DoorDash say about agent orders?
DoorDash reported that third-party agents account for only a few hundred orders in total. Xu said users still mostly place orders through DoorDash’s own app, according to TechCrunch. He framed the topic as a dynamic area and said the company does not want to ignore the trend or overstate it.
The source does not give a time period for the ‘several hundred’ figure, a breakdown by agent, or how many orders came from each agent. It is a single figure from an earnings call answer, not a audited metric, so readers should treat it as the company’s own estimate.
How is DoorDash building its own agent tools?
DoorDash is not waiting for outside agents to bring in orders. According to TechCrunch, the company has launched several products of its own:
- Ask DoorDash, a chatbot launched in June that lets users order through prompts and photos.
- A CLI agent introduced in July.
- A text-to-order agent launched in September.
- An MCP that lets other chatbots place orders for corporate ordering.
The MCP detail matters for developers. Model Context Protocol is an open standard for connecting AI assistants to external tools and data, so a company-released MCP lets a chatbot call an ordering service rather than scrape a website. The source does not say which chatbots support the MCP, what permissions it grants, or whether it is available to all businesses or only selected corporate customers.
Will DoorDash work with outside agent makers?
Yes, according to TechCrunch. Xu said DoorDash wants to learn more about agentic use cases and build a healthy relationship with third-party agent makers. He also said advertising inside those agents could become an opportunity in the future. The source gives no timeline or details for either plan.
The wider context is that websites are still deciding whether agents should reach their content at all. TechCrunch reports that Amazon has been blocking personal agents, while Walmart has welcomed them to win business from competitors. For a merchant or platform, that choice determines whether an agent can order at all.
Who it is for and where it falls short
This story is most useful to product managers, platform engineers and merchants who are deciding whether to build agent interfaces. The numbers suggest that agent ordering is not yet a large share of DoorDash’s business, so the case for heavy investment rests on future growth rather than current volume.
Developers should also note what the source leaves out. It does not name the agent makers’ products beyond the three names, does not explain how those agents connect to DoorDash, and does not give any usage trend over time. A single quarter’s comment cannot show direction. Without those details, the article cannot say whether ‘several hundred’ is growing, flat or falling.
Consumers and smaller restaurants may care more about the access question than the order count. If a platform blocks agents, users cannot order through them at all. If it opens access, the platform still controls the terms, and the source does not describe any such terms.
Bottom line: Treat the ‘several hundred’ figure as an early signal, not evidence that agent commerce is mature. If you build for DoorDash or a similar platform, watch for the MCP and advertising details, which the source does not yet specify.
Source and verification: This article is based on a TechCrunch report by Ivan, published on 30 October 2026, covering DoorDash’s third-quarter earnings call and comments from CEO Tony Xu. The 3 million daily deliveries figure comes from DoorDash’s 2025 report as cited by TechCrunch. The report contains no independently verified order data.
Related: datasette-llm 0.1a9 Now Uses LLM's Default Model · Zest's AI Restaurant Concierge Now Books Tables Over Text
Comments (0)