AI Was Supposed to Do Our Shopping. DoorDash Got Only Several Hundred Orders From It
DoorDash logged 993 million orders in the third quarter, yet outside AI agents placed only several hundred in total. Orders built through its own in-app AI ran about 50% larger.
Tell an AI, "Order me some chicken for dinner," and it picks the menu and pays. The technology is moving in that direction. But how many consumers actually use it? DoorDash's numbers were a surprise. The company had 993 million total orders in the quarter. Orders placed on shoppers' behalf by outside AI came to several hundred, cumulatively. Against the big story of AI remaking shopping, consumers were still tapping the apps they know.
1. 993 million vs. several hundred
On Oct. 29, DoorDash CEO Tony Xu disclosed an interesting pair of figures on the company's third-quarter 2026 earnings call.
| Category | Orders |
|---|---|
| Total orders, Q3 2026 | 993 million |
| Orders by outside AI agents | Several hundred (cumulative since launch) |
Source: DoorDash Q3 2026 results and conference call
Outside AI agents, including Meta's Muse and Instinct, have placed only several hundred orders on DoorDash for consumers since the service launched.
An AI agent is software that, given a goal by a person, carries out tasks such as searching, choosing and ordering.
The gap is clear. There is a wide distance between technology that can shop for you and consumers actually handing over the purchase.
For investors, the thing to read here is consumer habit, not AI performance. New technology does not change behavior overnight.
2. Does ordering chicken really need AI?
You open a delivery app on the way home. You pick the usual chicken and tap pay. A few touches and it is done.
Now hand the order to an AI. If you must answer "What kind of chicken do you want?", confirm the AI's pick, and check the price and delivery address, the task gets longer, not shorter.
If the task was easy to begin with, the reason to use AI weakens.
Compare planning a family trip. You weigh flights, check hotel locations and balance many conditions. Insurance and complex electronics purchases work the same way.
In those cases AI can cut the time spent searching and comparing. The success of AI shopping depends less on how flashy the technology is than on how much time and effort the consumer saves.
Seen that way, the forecast that AI will replace every shopping app looks different.
3. DoorDash's other number: baskets 50% larger
The same earnings report held a number pointing the other way.
Orders built through DoorDash's own AI feature, Ask DoorDash, had an average basket about 50% higher than conventional orders.
| Order method | Average basket index |
|---|---|
| Conventional order | 100 |
| Ask DoorDash | 150 (+50%) |
Why the difference? If AI understands a customer's tastes and situation and suggests fitting dishes or add-ons, the customer can build a fuller order.
An important distinction follows.
AI replacing the app and AI helping you shop inside the app are entirely different businesses.
The first is a contest to pull customers away from existing platforms. The second is a way for platforms to give customers more value. DoorDash's two numbers show exactly this difference.
4. What Amazon and Walmart should really fear
If AI takes over consumers' purchase decisions, who is hit hardest?
It is a key question for big retail platforms such as Amazon and Walmart.
Until now, shopping platforms earned influence as consumers visited, searched, saw ads and bought. If consumers start telling an AI, "Find the best product and order it," the starting point of the purchase changes.
The first party a customer meets is no longer the store but the AI.
Amazon's approach of strengthening AI inside its own shopping experience helps protect its direct relationship with customers. Walmart's competitiveness will depend on how it connects AI to its own customer experience and to outside purchase channels.
Advertising could change too. When people browse search results, ads influence the decision. When an AI compares products and recommends finalists, what product information enters the AI's judgment matters most.
Shopping ads may extend from a contest to buy human attention into a contest to enter the AI's recommendation process.
Investors should look less at the number of AI features and more at who secures the starting point of purchases and the final payment relationship.
5. The last barrier is trust, not technology
An AI recommending a meal is different from an AI paying with your card.
A bad recommendation can be ignored. A wrong order, a wrong address or an unwanted purchase is a problem.
For consumers to hand purchases to AI, three things are needed:
- Trust: Does the AI accurately understand my tastes and budget?
- Control: Can I limit the payment amount and product choices to a range I set?
- Accountability: Who fixes order errors, cancellations and refunds?
Companies that solve this can take a strong position in AI shopping.
Platforms such as DoorDash, Amazon and Walmart already run payments, delivery, refunds and customer service. Even if AI helps make the purchase decision, the ability to complete the actual transaction stays important.
Even in an era when AI picks the products, consumers want a platform that is accountable when something goes wrong.
6. Which numbers should investors watch?
Focusing only on user counts or new feature announcements could miss what matters.
These metrics are likely to connect more directly to company value:
| Key metric | Question for investors |
|---|---|
| AI agent order count | Do consumers actually give AI purchasing authority? |
| Conversion rate of AI users | Does AI turn browsing into payment? |
| Average order value | Do AI recommendations raise basket size? |
| Repeat purchase rate | Do customers who tried AI come back? |
| Profit per transaction | Does order growth turn into profit growth? |
For Amazon and Walmart, the question is whether AI can strengthen customer relationships and advertising competitiveness. For DoorDash, it is whether AI improves the ordering experience and raises transaction value.
AI agent companies face a different question: do they build a service convenient and trustworthy enough that consumers hand over payment authority?
The winner in AI shopping is likely to be the company that gives consumers a reason to come back, more than the one that adopted AI first.
7. INSIGHT TIMES VIEW: AI will likely remove the difficulty of shopping, not the shopping app
It is easy to see DoorDash's "several hundred" and conclude AI shopping is far off.
A slightly different reading is possible. The moments when shopping feels hard are not when we press the pay button. They are when we do not know what to buy, when we must choose among countless products, and when we worry about buying wrong.
That is where AI is likely to change things first.
For simple purchases like ordering chicken, existing apps remain convenient. For hard choices such as travel, insurance and high-priced goods, AI's role could grow quickly.
DoorDash's two numbers, several hundred and 50%, do not contradict each other. One shows how fast consumers hand over a whole purchase to AI. The other shows the value AI can add to an existing shopping experience.
The distinction matters for investors too.
There is no need to turn bearish on Amazon and Walmart simply because AI has arrived. Nor is there reason to be sure a newcomer will succeed just because it launched an AI shopping feature.
What matters next is not how many apps AI removes, but which consumer pain it removes and who captures the economic value.
The AI shopping revolution is likely to start less in a race to eliminate the pay button than in a race to own the moment a consumer decides to buy.
Insight Times Editorial Desk
