AI Agents Place Only Hundreds of Orders on Uber and DoorDash

Daniel Okafor

Uber handled 320 million ride-hailing and delivery trips during the last week of October, according to CFO Balaji Krishnamurthy, while AI agents supplied about 500 Uber Eats orders and four rides.

The disclosure came on Uber's third-quarter earnings call on Wednesday. Halloween orders helped make that week one of Uber's largest, Krishnamurthy said. He then set the boundary on adoption: “It is very, very early, and very limited traction that we have seen so far.”

How small is the agent channel?

Business Insider reported that Uber and DoorDash discussed hundreds of AI-agent orders on their latest earnings calls, including orders placed through systems such as Meta's Muse or Instinct. The totals remain small against the millions of rides and deliveries the companies complete each day, according to Business Insider.

DoorDash CEO Tony Xu said the company had received “several hundred orders in total” from AI agents. Xu made that disclosure a week before the Uber comments, Business Insider reported. The agent total remained a sliver of the billions of DoorDash orders delivered annually.

The companies' disclosed figures show the mismatch:

• Uber's last week of October: 320 million total ride-hailing and delivery trips; about 500 Uber Eats orders from AI agents; four agent-sourced rides.
• DoorDash: several hundred AI-agent orders in total, compared with billions of orders annually.

Predictions that agents would revolutionize online ordering have produced experiments, not material volume on the two platforms, Business Insider reported.

What are shoppers using agents for?

Some AI users have put agents in charge of buying everything from plane tickets to groceries, Business Insider reported. Shoppers have also said the tools are useful for monitoring prices, according to Business Insider's earlier reporting.

Business Insider previously reported that ordering errors and fears about agents going rogue with credit-card details have led others to use the tools more cautiously. Shoppers have said agents can make mistakes and need guardrails to be useful.

Business Insider's reporting identifies two barriers to wider use: agents sometimes make ordering mistakes, and users fear giving them control of credit-card details. Both concerns require controls before delegated purchasing can become routine.

What are the platforms building?

Business Insider reported that both Uber and DoorDash are moving ahead with AI tools on their own apps despite the limited usage already disclosed by the companies. DoorDash said in September that it was testing a text-to-order feature. The feature allows users to bypass the DoorDash app, for example by placing an order through text.

The practical change is access: users can reach DoorDash's ordering flow without operating its app, Business Insider reported. Neither company has framed agent traffic as a material share of orders in the cited disclosures.

What would count as traction?

The companies' disclosures describe agent ordering as a side channel, not a replacement for their main interfaces. For Uber, the 320 million-trip week set the scale; for DoorDash, billions of annual orders do.

The next test is whether first-party agent tools can move order counts from hundreds into thousands, and eventually into material shares of each platform's business. That requires tighter guardrails because errors and payment fears remain the barriers shoppers identified to Business Insider.

The next proof point is whether text-to-order and other native agent tools can turn hundreds of orders into sustained volume without repeating the mistakes and payment fears that limit current use.