DoorDash: AI agents sent only several hundred orders
DoorDash says third-party AI agents placed only several hundred orders in total. Agentic commerce is real but small. Here is how to budget for it.
DoorDash CEO Tony Xu told analysts that third-party AI agents have placed "maybe several hundred orders in total." Not per day. In total. DoorDash completes about 3 million deliveries a day. That is the most honest number on agentic commerce we have seen this year, and it comes from a company that has built more agent ordering tools than almost anyone.
What actually happened
On DoorDash's third-quarter 2026 earnings call, an analyst asked about orders that come in through AI agents. Per TechCrunch, Xu said agent ordering is "still a very dynamic area" and gave the several-hundred figure for third-party agents such as Muse, Instinct and Fo.
He also said "you never want to fight gravity," but "you don't want to be delusional." Xu said DoorDash wants a good relationship with agent makers, and that ads inside those agents could become an opportunity later.
DoorDash did not hold back on supply. It shipped a chatbot ("Ask DoorDash") in June, a CLI agent in July, a text-to-order agent in September, and an ordering connector built on MCP so other chatbots can place orders. The pipes exist. The demand from third-party agents does not, yet.
Why agentic commerce numbers matter for your business
Every vendor deck this year says AI agents will be your next big sales channel. DoorDash just gave a real denominator. Several hundred orders against roughly 3 million deliveries a day is a rounding error.
That does not mean you ignore agents. It means you size the spend correctly:
Do the cheap work now. Clean product and menu data, correct prices, structured attributes, and a readable site help agents, Google and humans at the same time. That work pays off today whatever agents do.
Don't fund a custom agent channel yet. If a vendor wants five figures to "make you agent-ready," ask for their order volume from agents. DoorDash's answer was several hundred.
Measure it yourself. Tag agent traffic and agent-placed orders in your analytics now. When the curve bends, you want to see it in your own data, not in someone's keynote.
Watch the ad angle. Xu named ads inside agents as a future opportunity. When agents do get volume, expect pay-to-be-recommended to follow, the same way search did.
Key takeaways
- DoorDash says third-party AI agents placed only "several hundred orders in total"
- DoorDash completes about 3 million deliveries a day, so agent volume is a rounding error for now
- DoorDash already ships a chatbot, a CLI agent, a text-to-order agent and an MCP ordering connector
- Fix your product and menu data now; it helps search and humans too
- Track agent-placed orders in your own analytics before you spend on an agent channel
Spend on the work that pays off with or without agents. We build clean catalog pipelines, structured product data and analytics that tag agent traffic, so you know when the channel is real. Run the numbers on your automation spend, or tell us what a vendor is pitching you.
Sources: TechCrunch, DoorDash Q3 2026 results, DoorDash ordering connector.
- #doordash
- #ai-agents
- #agentic-commerce
- #restaurants
- #mcp
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
AI training opt-out: one shared robots.txt token for AI bots
Anthropic, OpenAI, Google and Microsoft back a shared robots.txt AI training opt-out token. Here's the syntax, the dates, and what it won't stop.
Read itZest AI restaurant agent books tables over text: OpenTable, Resy
Zest launched an AI restaurant concierge that books via OpenTable, Resy and SevenRooms over text. Your reservation system is now an agent channel.
Read it