DoorDash iMessage AI agent: your menu data is the order
DoorDash opened a US waitlist for a text-to-order AI agent in Apple Messages. Early tests showed wrong prices. Here is what restaurants should check now.
DoorDash opened a US waitlist on September 30 for a text-to-order AI agent that runs inside Apple Messages. A customer texts "order my usual," and the agent builds the cart, picks the restaurant and charges the card on file. No app, no menu scroll, no storefront. For a restaurant, that means the customer may never see your menu page. The agent reads your menu data and decides what to say about it.
What actually happened
Per TechCrunch, the DoorDash AI agent can:
- Repeat recurring orders ("order my usual")
- Suggest dishes and nearby restaurants, and send food photos with the suggestion
- Build group orders with mixed dietary preferences and different quantities in one cart
The agent matches the texter's phone number to a DoorDash account. That gives it order history, preferences and the stored payment method. Access is a US waitlist. DoorDash has not given a date for general launch.
The early build had problems. Bloomberg tested it and, as The Next Web reports, found prices that differed from the app and photos that did not match the recommended item. DoorDash said waitlist users will get a newer version.
Some outlets report a specific pilot user count. We could not confirm that number against a primary source, so we leave it out.
Why a text-to-order agent matters for your business
A text thread has no room for a hero image, a combo upsell banner or a loyalty popup. The agent picks one or two items and one photo. Three things decide whether it picks yours:
Your menu data must be correct. Bloomberg saw price mismatches in a DoorDash-built agent. If a price, modifier or item photo is wrong in your merchant portal, the agent can repeat the error to a customer who has no menu to check. Audit prices, modifiers and photos on every delivery platform this week.
Dietary tags now sell food. The agent builds group orders around dietary needs. An item without a "vegetarian" or "gluten-free" tag can drop out of that cart. Tag everything that honestly qualifies.
"My usual" favors the incumbent. Repeat orders are the easiest thing for the agent to do. Your first order from a customer now matters more, because the second one may happen with no browsing at all.
Also keep a direct channel. When the order flows through someone else's agent, the customer relationship belongs to that agent.
Key takeaways
- DoorDash opened a US waitlist on September 30 for an AI agent that takes food orders in Apple Messages
- It uses the customer's order history and stored payment method, and handles group and dietary orders
- Bloomberg's test of an early version found wrong prices and mismatched photos
- Audit your menu prices, modifiers, photos and dietary tags on every delivery platform
- Keep a direct ordering channel so the agent is not your only customer relationship
When an agent reads your menu, bad data becomes a wrong order. We build menu and catalog pipelines that keep one source of truth in sync across delivery platforms and your own ordering site. See what we build for restaurants and retail, or send us your menu sync problem.
Sources: TechCrunch, Bloomberg, The Next Web.
- #doordash
- #ai-agents
- #restaurants
- #conversational-commerce
- #menu-data
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
Braze Conversational Agents sell over WhatsApp and SMS
Braze put Conversational Agents for WhatsApp, SMS, RCS and web into beta, plus an MCP layer for Claude and ChatGPT. Here is what to feed a sales agent.
Read itManus Cue gives every agent an email, phone, and wallet
Manus 2.0 shipped Cue: personal AI agents with their own phone number, email address, and budgeted wallet. Your phone line is about to take bot calls.
Read it