Uber's Product Chief on Hotels, Robotaxis, and Why Less Is More

Over the past year, Uber has quietly expanded its operations beyond its widely recognized ride-hailing and delivery services. While these remain core, the Uber application now integrates hotel bookings powered by Expedia, offers "shop for me" concierge features, and facilitates boat rentals in Europe, signaling a broader strategic vision.

Internally, significant developments are also underway. These include the introduction of debit cards for drivers, a data-labeling initiative providing additional income opportunities for earners, and the establishment of AV Labs. This six-month-old business unit is dedicated to developing a fleet of sensor-equipped vehicles, distinct from Uber's standard driver network, to accumulate extensive driving data. Uber presents this initiative as a means to strengthen relationships with its autonomous vehicle partners, some of which it holds equity in. However, it also appears to serve as a strategic hedge; Uber directly competes with several of these partners, notably Waymo, and controlling the data layer grants Uber both leverage and future optionality.

The question of whether Uber will evolve into a comprehensive "everything app," akin to Asian super-apps like Grab, remains to be seen. In a recent conversation, Uber Chief Product Officer Sachin Kansal provided TechCrunch with insights into the company’s financial services ambitions, its increasingly complex relationship with Waymo, the new AV Labs data operation, and how artificial intelligence is beginning to manifest in ways riders and drivers will tangibly experience.

This interview has been edited for length and clarity.

TC: Earlier this year, Uber unveiled hotels, boat rentals, and expanded shopping features. How was this selection made, and what initiatives did not proceed?

SK: Annually, our teams develop numerous initiatives, and a select subset is chosen for a prominent public launch. This year, the overarching theme we focused on was travel. With 1.5 billion trips on the Uber platform occurring outside a user's home city each year, we recognize travel as a prevalent use case for our users. Our headline announcement this time was the introduction of hotels on Uber, facilitated through a partnership with Expedia. However, travel encompasses much more—it involves rides from the airport to the hotel, and food. We observed that many users had shifted from room service to using the Uber Eats app. With "shop for me," our objective was to enable users to shop from any local store, even if its full catalog isn't available on Uber Eats. In my view, travel represents the third pillar of our service, following rides and eats.

Is Uber progressing towards offering its own financial services, similar to the "everything apps" prevalent in Asia?

Our approach to financial services spans various entities: consumers, drivers and couriers, and merchants. Currently, we offer multiple products primarily for drivers and couriers, such as the Uber Pro card, which functions as a debit card for transferring earnings. We are also beginning to pilot similar products for merchants in specific global regions. Regarding consumer financial services, we are still evaluating its long-term viability. Presently, consumers utilize Uber credits, which are linked to our membership program. For instance, Uber One members receive 10% cash back on hotel transactions, meaning a $1,000 booking yields $100 in credits usable for rides and eats.

Would Uber ever introduce its own buy now, pay later product?

I am uncertain about that, as we prefer to collaborate with industry experts in such specialized areas. We have already announced partnerships with third-party providers who offer this service, allowing users to utilize it at checkout. Our general product strategy is not to be everything to everyone.

With boat rentals in Europe, tapping the tab directs users to a partner's booking flow rather than completing the transaction within Uber. Is this handoff model a template for future integrations?

Certainly, in certain instances, particularly when launching new services, relying on partners is beneficial. A deep two-way integration requires considerable time, and sometimes it's prudent to test a concept before fully integrating. In the case of Expedia, we decided deep integration made sense, and we built the entire user interface ourselves in collaboration with them. However, in other scenarios, it may be more appropriate to hand off the remainder of the experience to specialists in that field. If significant traction is achieved, we can always pursue deeper integration later.

Your Uber One membership product now boasts 51 million members, accounting for approximately half of all bookings. Do you have data demonstrating effective cross-selling—for instance, a delivery user subsequently taking more rides?

On the delivery side, members typically break even on their monthly fee within two to three orders. As members become more accustomed to the program, we observe an increase in their frequency within their primary line of business. Furthermore, it encourages greater usage of other services—we see mobility-only users starting to utilize delivery, and delivery-only users beginning to use mobility.

Delivery has historically been one of the most challenging tech businesses to make profitable. Is Uber Eats still reliant on ride-hailing for its financial health?

During its initial years, Uber Eats was not profitable. However, over the last several quarters, Uber Eats has independently become a profitable business for us, generating substantial profit.

A story I wrote this spring suggested Uber is increasingly competing directly with Airbnb, which now offers airport transfers through a partner. Do you perceive it this way? Who are your primary focus areas?

There is an abundance of competitors—Lyft in the U.S., Didi and 99 in Latin America, Bolt, Ola globally, and DoorDash, Delivery Hero in the delivery sector. However, I dedicate a very small portion of my time to considering them. The greater percentage of my time, and what primarily concerns me, is ensuring we provide all possible value to our users.

You recently wound down the Waymo pilot in Phoenix while expanding elsewhere. How do you maintain a coherent user experience when partnering with—and in some cities competing with—the same supplier?

Phoenix was our initial launch city with Waymo, deploying about a dozen vehicles. Our larger-scale launches have been in Austin and Atlanta, where we operate hundreds of vehicles with them. When we recently reviewed the Phoenix pilot, we mutually concluded that its continuation was not strategically sound. Waymo is an excellent partner, but in many cities, they are also a competitor. We are not aiming to be an L4 autonomy provider; our focus is on establishing the infrastructure to collaborate with multiple players. We advocate for a hybrid network—human drivers alongside autonomous vehicles in the same city—as it allows us to effectively balance demand and supply.

Regarding AV Labs, what unique offerings can Uber provide to autonomy partners?

We will be equipping hundreds of vehicles with sensors, deployed through our fleet partners, and through this, we will collect millions of miles of driving data. This is crucial for addressing the "long-tail problem"—capturing all edge cases, not just the P95 or P99 level. Beyond the data itself, we offer extensive operational expertise derived from our 10 million earners, particularly regarding pickups and drop-offs. We manage 25 million lost items annually—understanding how to operationally handle such scenarios in an autonomous world is the kind of expertise we bring.

Is Uber selling driver and rider data to Generative AI companies?

I would categorize this into two parts. Regarding Generative AI companies, we utilize our earner base for data labeling, including audio collection, and yes, we have commercial agreements to sell this data to them. This represents a new business segment that we are extremely optimistic about. AV Labs is a separate initiative, and we are still developing the models for sharing that data with partners; it's still in the early stages.

Are drivers recording conversations with riders for this data work?

No, absolutely not—I want to be very clear that no conversations are recorded as part of this initiative while drivers are on a ride. When they are off-trip, not driving or delivering, they might be speaking, or listening to an audio segment and transcribing it. They are compensated for this work.

Where has AI genuinely manifested in ways a rider or driver would notice?

For earners on our platform, we have an earner assistant. Their primary concern is maximizing earnings, and the assistant might advise, "It's relatively quiet in the South Bay, but you might find significant demand five miles away." On the Eats side, a grocery cart assistant allows users to say, "I want milk, eggs, bread," and it quickly compiles the cart. For rides, users can use voice commands to request a ride, for example, "I'm looking for a ride to the airport, I have six pieces of luggage, six people."

So, is a fully agentic Uber—where the platform "plans and books my whole trip"—on the horizon?

I cannot provide a specific timeline, nor can I detail the exact feature set, but I believe AI will be a significant enabler for such an experience. It would allow users to delegate complexity to the platform and simply communicate their desires to an agent. This is easier said than done; we aim to ensure we're not merely checking a box by deploying an agent that doesn't perform effectively.

As CPO, how do you personally prioritize with so many ideas in development?

I would say I dedicate 70% to 80% of my time to ensuring our existing products, or those nearing launch, are as robust as possible. New ideas, while often captivating, are like novel concepts—out of a hundred, perhaps five are truly good, and those five then require extensive cultivation and conviction. So, approximately 20% of my time is spent on new ideas. This includes, by the way, personally driving and delivering to gain firsthand insight into our product from the other side.