What happened
On Uber's third-quarter earnings call on November 4, 2026, its investor relations head, Balaji Krishnamurthy, said Uber handled about 320 million trips and orders across Uber and Uber Eats in Halloween week, one of its biggest weeks. AI agents brought in around 500 Uber Eats orders and, in his words, "a grand total of four trips" for rides. CEO Dara Khosrowshahi called it "very, very early and very limited traction." The Verge says it was the first time Uber had disclosed these figures. Khosrowshahi also said Uber is talking to AI assistant companies and "negotiating the terms of trade," and that it wants to avoid depending on any one player. He pointed out that booking a ride in the Uber app takes a median of about 50 seconds. Separately, Skift reported on November 9 that Meta's Muse agent can now book Expedia hotels in the US, about six weeks after the two companies announced the integration. Expedia told Skift it covers natural-language search and booking but not cancellations or rebooking. Because Expedia is the merchant of record, changes still go through Expedia's customer support. And on November 5, Airbnb's shareholder letter said its own AI travel agent will launch in the first half of 2027 and will help guests after booking, with changes, payment issues, refunds and safety support, which Airbnb says an outside agent cannot easily do on its own.
Why it matters for business owners
Most advice about customer AI agents is written as if the traffic has already arrived. Uber is a company that could see it first, and it counted four rides. That is a useful correction for an owner who is being pitched agent-ready websites, agent checkout work, or new protocols. Those may become worth doing. The question is when, and for which part of your business. The Expedia detail matters just as much. Even in a live integration, the agent starts the sale and the business still handles everything that goes wrong afterward. Changes, cancellations and complaints land on your team, whoever found the customer.
What owners should not misunderstand
Do not read four rides as proof that agents will never matter. Uber itself says it hopes for more traction and is negotiating with the assistant companies. This is an early reading from one very large consumer app. Do not assume your numbers look like Uber's. A hotel booking, a B2B quote, a home service request and a food order are different purchases, and agents will reach them at different speeds. Uber's customers already have a very fast direct path. Yours might not. Do not assume an agent integration means you give up the customer relationship. The Expedia example shows the opposite for now: the business remains the seller of record and the support desk. Do not confuse this with using AI inside your company. Airbnb reports that about 40% of its support issues were resolved without a human agent in the third quarter, up from a third in the second, and that is Airbnb's own statement about its own assistant. Customer-facing agents from outside companies are a separate question from the AI you run yourself.
The operational lesson
Treat agent traffic like any other new sales channel. You would not rebuild your quoting process for a channel that sent you four leads, and you would not ignore it if it started sending four hundred. The practical move is cheap: find out whether it is happening at all. Ask your web host, your booking or ordering software, and your phone or chat provider whether they can tell automated assistant traffic from human traffic. Many cannot yet. Then decide what you would allow before anyone asks. Outside agents may be fine for browsing prices and availability and risky for refunds, account changes or anything that moves money. Write that down once, so the first request does not set your policy by accident.
What a serious business should do next
Check your own data. Look at the last 90 days of orders, bookings or leads and ask your software vendors whether any came through an AI assistant. If nobody can answer, that is the finding. List the customer actions you offer and mark each one: fine for an outside agent, human only, or unsure. Start with the ones that involve payment, personal data or a promise your team has to keep. Ask vendors in writing what they plan. If your website, booking or commerce platform adds agent support, find out what it costs, what data it exposes, and whether you can turn it off. Keep your direct path fast. Uber's argument is that its own app is quick enough that customers will not need a middleman. For many small businesses, a clear page, a short form and a reply that comes the same day does more than any agent integration. Set a date to look again. A quarterly check of your own numbers is enough. If agent-originated orders show up in real volume, you will have a baseline, and the decision gets easy. When not to act: if your business sells mostly by relationship, quote or referral, agent preparation is unlikely to be your best AI spend this year.
The Atlacis view
Atlacis helps owners slow down, understand the workflow, map the risk, choose the right tool or system, and avoid wasting money on AI decisions that do not match the business. Preparing for a channel that has not shown up in your numbers is a good example of a decision that deserves a measurement before a budget. If you are being pitched agent readiness and want a second opinion on what is worth doing now and what can wait, a short call can help you sort it.
The short version
- Uber said AI agents fulfilled four rides and around 500 Uber Eats orders in a week of roughly 320 million trips and orders.
- Expedia's Muse integration covers search and booking, not cancellations or rebooking, and Expedia remains merchant of record.
- Airbnb says its own AI travel agent is due in the first half of 2027.
- Check your own order and lead data for assistant traffic before paying for agent readiness.
- Decide in advance which customer actions an outside agent may take, and keep your direct path fast.
Where ATLACIS can help
Sources
- The Verge: Uber only fulfilled 4 rides from AI agents last quarter (November 4, 2026)
- TechCrunch: Almost no one uses AI agents to book their Uber, company says (November 4, 2026)
- Skift: Expedia Goes Live on Muse, but Its AI Wins Are Inside Its Own Apps (November 9, 2026)
- Airbnb: Q3 2026 financial results (November 5, 2026)