DoorDash opened a door to third-party AI ordering less than a month ago through its Model Context Protocol connector, a link designed to let outside tools place orders on the platform. The rollout remains limited to early beta partners and has been used chiefly to streamline corporate lunch orders.¹

The first results are stark. On the company’s third-quarter earnings call, co-founder and CEO Tony Xu said outside agents had generated only “several hundred orders” since becoming available — a microscopic total beside the nearly 1 billion orders DoorDash handled during the quarter.¹

Xu’s explanation draws a line between the chores AI might eventually take over and the ones consumers still find easy enough to do themselves. Agents could prove useful for high-consideration purchases or decisions requiring background research, he said. But “when it comes to a task that’s as simple as ordering food, we tend to see that come directly to DoorDash.”¹

That does not mean DoorDash is stepping back from AI. Its own Ask DoorDash feature can build a cart from a recipe photo or make suggestions for a loosely defined meal. The company has said baskets built through the tool are created five times faster, carry nearly 50% higher value and include about 60% more unique items than customers’ conventional orders; almost half go to local restaurants the customer has not previously tried.¹

The contrast is telling. External agents have barely registered, while DoorDash’s embedded tools are being pitched as a faster route to discovery rather than a substitute for customer choice. Xu’s verdict on the broader technology was cautious: “you never want to fight gravity,” but “you don’t want to be delusional.”¹ For now, the gravitational pull is toward the familiar app.