Uber Handled About 320 Million Trips and Orders in a Week. AI Agents Booked Four Rides.

Four rides — that's how many rides came from an AI agent during the company's busiest week of October. Across Uber and Uber Eats, the company handled roughly 320 million trips and orders that week. AI agents placed 500 Uber Eats orders.

Halloween week: four AI rides out of 320 million trips and orders

Uber reported the numbers on its third-quarter earnings call on Wednesday. During Halloween week, its busiest stretch in October, just four rides were ordered by an AI agent.

Across Uber and Uber Eats, the company said it handled roughly 320 million trips and orders that week. AI agents placed 500 Uber Eats orders.

Balaji Krishnamurthy, Uber's chief financial officer, gave Wall Street analysts those numbers on the call. CEO Dara Khosrowshahi called the numbers "very, very early and very limited traction that we are seeing so far". (TechCrunch)

It isn't just Uber. DoorDash said on its earnings call last week that AI agents have sent it just a few hundred orders so far, since most customers still order in the app.

So if the agent takeover already happened, somebody forgot to tell the customers. And that gap is exactly where your sales floor lives right now.

The money says the opposite of the numbers

Here's the strange part. A wave of personal AI startups has arrived.

Instinct, a personal AI startup founded last year, raised its most recent round at a $10 billion valuation. Meta's Muse hit No. 1 among U.S. apps in Apple's App Store last month and passed 10 million users, by Meta's own reporting, and the company said many people were using it to shop and buy from businesses.

OpenAI, Anthropic, Google and SpaceXAI all offer agents now, along with many smaller startups.

Khosrowshahi said the market has about eight major AI assistants today, and he expects the number to pass 10.

A $10 billion valuation on one side. Four rides on the other. Something has to explain that.

Fifty seconds is the whole story

Booking a ride in Uber's app takes 50 seconds on average, and re-ordering a favorite meal takes 90 seconds. TechCrunch's take: an agent may not save the buyer time.

That's the whole thing in one number. My read: your customer doesn't want the smartest path to a purchase. They want the shortest one. Every app on their phone has trained them for years to expect exactly that.

And Khosrowshahi isn't writing agents off. He said, "We think that is another opportunity to reach our platform." He also said, "Listen, we hope that there will be more traction." (TechCrunch)

My read: he's playing a long game. Which raises the question every owner should be asking: if the agent wave is still this early, what's eating your leads today?

What it means for an owner who sells (my read)

This part is my read, not reporting. Here's how I'd think about it in the industries I spend the most time around.

Golf cart dealers. Picture a dealer with a lot full of carts and a form on the website. A retiree fills it out late one night asking about a cart for the neighborhood. Nobody calls him back until days later. By then he's driven to the dealer across town and put a deposit down on a cart that isn't even the one he wanted.

Nothing in that story involves an AI agent. It involves a long gap. Uber's 50-second number is a useful reminder of what fast looks like right now — your buyer has been trained by every app on their phone to expect an answer now, not days later.

Roofing contractors. Picture a storm rolling through. A homeowner's ceiling stains. She searches, taps a few quote buttons, and calls the first number that picks up. The roofer who answers books the inspection. The others are texting her days later, when she's already signed.

My read: the agent story is a distraction from the boring problem. Your competitor isn't an AI. It's the guy who answers his phone, and the shop that follows up before the lead cools off.

So what do you actually do about it this week?

How to use it this week

1. Count your dead leads. Open your lead list and pull every inquiry from the last month that got one touch and nothing after. Write the number down. Then compare it to the number of jobs you closed. It takes one sitting.

1. Run the after-hours test. Have someone you trust fill out your website form after dinner on a weekday and time how long it takes for a human to respond. Not an auto-reply — a human. Uber's 50 seconds is the bar. Time yours and see where it lands.

1. Write your first-touch text for each line of business. Keep it short, ask one question, offer one next step. Here's a golf cart version:

Hi [name], this is [you] at [dealership]. You asked about the [model]. Quick question so I point you right: is this for the course or around the neighborhood? I can send photos of what's in stock today.

1. Write the roofing version. Same rules, different question. Roofing leads want to know one thing first: is somebody coming, and when.

Hi [name], [you] with [company]. Got your request about the roof. Is water coming in right now, or is this for a quote? Either way I can get someone out this week — what's the address?

1. Write one follow-up for later in the week. My read: the second touch is where the money is. This is the cheapest money in your business.

Hi [name], following up on the [project]. Is it still an issue, and do you want the estimate by text or in person? I have an opening later this week.

1. Put one person on the clock. Whoever owns the phone owns the lead. If a lead sits past a full day with no human touch, it goes to whoever's next in line. No exceptions.

1. Re-run the count next week. Same list, same time window. If the dead-lead number didn't drop, your problem isn't leads. It's the follow-up.

Mistakes to avoid

• Don't buy an AI agent because a valuation headline scared you. Uber's own CEO says the traction is early and limited. You have time.
• Don't let a bot be the last thing a lead hears. An auto-reply is not a touch. A human asking a question is.
• Don't measure your response time on your best day. The after-hours test exists because that's when real leads come in.
• Don't assume forms are dead. Uber's numbers say people still use the app they already have. Your customer still uses the form they already found.