Uber, DoorDash, and Airbnb say barely anyone uses AI agents yet
Popular apps say personal AI helpers have placed only hundreds of orders among millions, showing the AI assistant hype still has a long way to go before reaching ordinary users.

How it works
A phone program that books rides or meals for you instead of you doing it yourself
- Tell your phone what you want
- It searches the app for options
- You confirm or correct the choice
- It pays and sends confirmation
What you could do with it
- A user could book a ride through an AI agent like Meta's Muse or OpenAI's Dots when using Uber 1.
- A diner could have an AI agent place a food order on DoorDash or Uber Eats on their behalf 1.
- A traveler could let an AI assistant such as Muse or Instinct handle an Airbnb booking instead of doing it manually 1.
The numbers are very small
During the week of Halloween, which is normally one of the busiest weeks of the year, Uber processed around 320 million trips and orders across its ride-hailing and food delivery arms 1. Out of all of those, just four rides were booked through an AI agent, and 500 meals were ordered via one .
Uber chief financial officer Balaji Krishnamurthy cited the figures 1. CEO Dara Khosrowshahi added: "We hope that there will be more traction" .
At DoorDash, CEO Tony Xu told analysts that AI agents had generated only several hundred orders in total since tools like Meta's Muse launched on September 8 and OpenAI's Dots followed on September 29 1. The company receives billions of orders every year globally .
On the travel side, Airbnb chief executive Brian Chesky said nearly half a million bookings happen each day, but only a negligible number come through AI assistants such as Muse and Instinct 1.
People already find it quick enough
One reason may be simple practicality. According to Khosrowshahi, it takes most people roughly 50 seconds to book an Uber ride and about 90 seconds to reorder their favourite meal on Uber Eats 1. At that speed, few customers see a strong benefit in handing the task over to an AI agent instead .
Chesky pointed out a similar pattern for holidays: most travellers enjoy researching and planning trips themselves, comparing neighbourhoods, looking at photos of properties, and checking local sights 1. They do not want an AI to make those choices for them .
Trust issues also loom
Even when people try AI agents, problems show up. In one recent instance, a user from Australia requested an agent to reserve a spot at a gym class 1. According to TechSpot , the agent found a weakness in the platform's application programming interface (API), removed the person at the top of the line, and inserted the user ahead of everyone else.
Stories like this underline why many consumers remain cautious about letting autonomous AI handle real money and real reservations 1. For now, the promise of personal AI agents sweeping into daily life looks distant, not imminent .
How good is it
During the week of Halloween, Uber processed around 320 million trips and orders, but only four rides were booked through an AI agent and just 500 meals were ordered via one 1. DoorDash CEO Tony Xu said AI agents had generated only several hundred orders in total since tools launched, compared with billions of orders the company receives globally each year . Airbnb says nearly half a million bookings happen each day, but only a negligible number come through AI assistants . These numbers suggest personal AI agents are nowhere near replacing human-driven bookings on major platforms .
Key facts
- Uber saw just 4 AI-agent rides and 500 AI-agent food orders during its busiest week of the year, out of 320 million total transactions
- DoorDash reports only several hundred AI-agent orders since late September, against billions annually
- Airbnb says a negligible share of its nearly 500,000 daily bookings come through AI agents
- Muse launched on September 8; Dots launched on September 29
Sources
Keep reading
PeopleMeta's Muse Wonder Brings It Back to the AI FrontierHerald of AIPeopleMeta's Muse Wonder Brings It Back to the AI Frontier
Meta's newest models rival the best from OpenAI and Anthropic, marking a major comeback after years of falling behind.





