AI Shopping Agents Have Placed Only Several Hundred DoorDash Orders. Ever.
Quick Take
- DoorDash CEO Tony Xu said third-party AI agents have produced only “several hundred” orders so far, per reports on the earnings call.
- In the same quarter, DoorDash handled 993 million total orders, according to CNBC.
- DoorDash’s own in-app AI tools are a separate story, and the company says those orders have bigger baskets.
- Early days, yes. But the gap between agent-commerce hype and real use is huge.
“Let the AI order dinner” is one of the most repeated promises in the agent pitch. DoorDash just put a number on it, and the number is tiny.
What DoorDash said
On its third-quarter earnings call, an analyst asked about AI agents. Per TechCrunch and Ars Technica, Xu said agents from outside companies have sent maybe several hundred orders in total since they started. TechCrunch names a few of those agents, including Muse and Instinct. Ars reports he called it a very low use case.
Xu also said you should not fight where technology is heading, but you also should not be delusional about it, according to both outlets. He added that for a task as simple as ordering food, people tend to come straight to DoorDash.
The scale gap
CNBC reports that DoorDash handled 993 million total orders in the quarter, up 28% from a year earlier. The total value of orders on its platform grew 36% to $33.9 billion, and revenue came in at $4.62 billion. Ars puts it more simply: just under a billion orders in one quarter.
Put those together. Even if “several hundred” meant 900, outside agents would equal less than one in a million of the orders from this quarter alone. And the agent number covers their whole lifetime, not just one quarter.
What DoorDash is building anyway
The company is not ignoring AI. TechCrunch lists its own tools: an Ask DoorDash chatbot from June that takes orders from prompts and photos, a command line agent from July, a text-to-order agent from September, and an MCP connection that lets other chatbots place corporate orders.
Ars notes that the MCP connector has been live for less than a month and only for a small group of beta partners. So outside agents have had very little runway. Without that link, an agent has to fight through the regular website.
DoorDash also says its own Ask DoorDash orders have bigger baskets than users’ usual orders, per Ars. That is a company claim, and it is about DoorDash’s in-app assistant, not outside agents.
The strategy tell
In the same week, DoorDash agreed to buy Instacart for about $17 billion, or $65 a share in cash and stock, per CNBC. That is a huge bet on grocery delivery, drivers and store partners. The company is spending its biggest money on the physical side of shopping, not on the AI assistant that might sit in front of shoppers.
Fact, claim, judgment
Fact: outside AI agents account for a vanishingly small share of DoorDash orders today.
Claim: that agents could later matter for bigger purchases where research pays off, and maybe as an ad channel. That is Xu’s view, as reported, not a result.
Judgment: this is the cleanest reality check on agent shopping we have seen. Ordering food is quick and habitual, and the app already does it well. The counterpoint is fair: the official agent connection barely exists yet. Check back in a year.
What you can do
When a company claims agents are “transforming” shopping, ask for its order counts. DoorDash shared one, and it is small.
Bandwagon Check
Fairly hyped as a long-term bet that agents will one day handle bigger, research-heavy purchases; overhyped as proof that agents are already changing how people shop. At one of the biggest delivery platforms, outside agents have produced a few hundred orders against nearly a billion in a single quarter. The official connection is new, but the hype is far ahead of the habit.
Sources
- TechCrunch: DoorDash said it’s only received several hundred orders from AI agents
- Ars Technica: Basically no one is using AI agents for their DoorDash orders
- CNBC: DoorDash to acquire Instacart for $17 billion
By Willa Ravenwire · AI Bandwagon
