DoorDash CEO Tony Xu says third-party AI agents have placed several hundred orders so far, a cumulative figure not a Q3 total.
DoorDash CEO Tony Xu said third-party AI agents have generated only “several hundred” orders on the delivery platform so far. He made the disclosure during the company’s
October 29 earnings call, as reported by TechCrunch, describing outside-agent use as low. The figure is cumulative since the agents began operating, not a count for the third quarter alone. It offers an early reality check on claims that AI assistants are already changing how people shop.
What DoorDash said about AI agent orders
Xu answered an analyst’s question about AI agents during DoorDash’s third-quarter 2026 call. “AI agents are generating several hundred orders across the time that they’ve been up and running.” — Tony Xu, DoorDash co-founder and CEO. He called the use case “fairly low” and said most users still place orders directly through DoorDash.
The disclosure concerns agents made by outside providers, not every AI feature DoorDash offers. TechCrunch identified third-party agents including Muse, Instinct and Fo. Xu said DoorDash wants to understand this developing area and maintain a constructive relationship with agent makers.
The number should not be read as an exact adoption rate. DoorDash gave a rounded total and did not specify how many agents were included, how long each had been available or what counted as an agent order. The company also did not break out how many of those orders came through a particular integration.
A small count beside DoorDash’s scale
DoorDash reported 993 million total orders in the third quarter, up 28% from a year earlier. Its
official results release also recorded $33.9 billion in Marketplace gross order value and $4.6 billion in revenue. These figures show the scale of its business, but they are not directly comparable to the several-hundred cumulative agent orders.
The time periods differ. The 993 million figure covers only three months, while Xu described agent orders over the period those agents had been running. Even so, the comparison indicates that outside agents remain a very small channel relative to DoorDash’s main ordering business.
Xu said ordering a meal is a relatively simple task that customers often bring straight to DoorDash. He suggested agents may be more useful for longer, more involved decisions, where they can work proactively in the background. That is the CEO’s view of potential use cases, not evidence that consumers are already adopting agents for those tasks.
DoorDash’s own AI ordering tools
DoorDash is also developing its own AI features and agent integrations. The company launched Ask DoorDash, a consumer assistant for discovering restaurants and building grocery baskets. It has also introduced text-based ordering and a command-line tool for developers.
In September, DoorDash announced a 
Model Context Protocol connector for corporate ordering. The connector lets compatible workplace AI tools search items, create a cart, place an order and track delivery. DoorDash said access would begin with a limited beta for companies ordering for employees.
That business-to-business pilot should be distinguished from mass consumer access. The connector was designed for office lunches and replenishing workplace supplies, and beta access was limited. 
Ars Technica reported that only a small group of early partners could use it, so the current order count reflects an early, constrained rollout rather than a mature product.
What the agent-order figure can tell us
The disclosure provides a snapshot of actual orders, not a forecast for the entire agent-commerce market. It suggests that, at least on DoorDash so far, outside AI assistants have not displaced the company’s own app as a major ordering route. Xu said the company is watching the space rather than dismissing it.
Several factors could affect how the number changes. More agents may become available, integrations may become easier, and users may find new tasks they want assistants to handle. At the same time, customers may prefer familiar apps for routine purchases, especially when they already have saved preferences and payment details.
Ultimately, DoorDash’s disclosure tempers near-term claims about agent-led shopping without settling what comes next. The company has begun testing ways for agents to place orders, while the current volume remains small by its own description. Wider access and more time will show whether these tools become a meaningful route to local commerce.
