Uber Technologies Is Teaching Your Phone to Grocery Shop—And That’s the Point

What changed this week

Uber Technologies, Inc. (UBER) just gave Uber Eats a new trick: an AI feature called “Cart Assistant” that can build a grocery cart from a text prompt or even an image—think a photo of your handwritten list, a recipe screenshot, or “snacks for movie night.” The company rolled it out in beta in the U.S. on February 11, 2026, positioning it as a time-saver and, more importantly, a habit-builder.

This isn’t Uber trying to be cute with AI. It’s Uber pushing deeper into the most valuable place a consumer app can live: the weekly routine.

From “ride app” to “life app,” one errand at a time

Uber’s pitch has been consistent for years: open one app and solve the annoying parts of modern life—getting across town, getting dinner, getting groceries, getting a package across the city. The challenge is that rides are frequent, but not always daily; restaurant delivery is addictive, but also easy to churn; groceries are both high-frequency and high-stakes (you notice immediately when something is wrong).

Cart Assistant is designed to reduce friction at the exact moment people abandon a grocery order: the blank-cart problem. Staring at an empty cart feels like work. Uploading a list and letting the app do the first pass feels like outsourcing. Uber is betting that “outsourcing” becomes a default behavior.

It also nudges Uber Eats into more “agent-like” behavior—less browsing, more doing. That’s the direction consumer tech is heading in 2026: fewer endless menus, more systems that take a request and return a completed task you can edit.

The bigger story: Uber’s platform is throwing off real cash

Here’s what makes the AI grocery feature more than a PR beat: Uber is rolling into 2026 with scale and momentum that looks increasingly durable.

In its February 4, 2026 results for the fourth quarter of 2025 (ended December 31, 2025), Uber reported:

• 3.8 billion trips in Q4 2025, up 22% year over year
• $54.1 billion in gross bookings in Q4 2025, up 22% year over year
• $14.4 billion in revenue in Q4 2025, up 20% year over year
• $2.8 billion in free cash flow in Q4 2025
• $10.0 billion in free cash flow for full-year 2025

If you’ve followed Uber for a while, you know why this matters: for years, the company was defined by growth at all costs, regulatory drama, and the ongoing question of whether ridesharing could be sustainably profitable. Now the narrative has shifted toward a platform that can fund its own experiments—AI features, new delivery categories, and partnerships—without constantly asking markets for patience.

Robotaxis: Uber’s not trying to win the lab race

Autonomous vehicles are the looming “what if” over every ride-hailing model. Uber’s current posture is basically: we don’t need to build the car, we need to be the network.

The company has been explicit that it wants to be a major facilitator of autonomous trips, leaning into partnerships rather than trying to out-invent dedicated AV players. In plain language: if robotaxis are real and scalable, Uber wants them plugged into the same demand engine that already moves tens of millions of rides per day.

That’s the connective tissue between Cart Assistant and robotaxis, oddly enough. Uber isn’t selling a single product; it’s selling repetition. The more often you open the app, the more natural it feels to use Uber for the next thing.

What to watch next

Cart Assistant is launching in beta, which means it’s also a test of trust. Grocery orders are less forgiving than takeout: substitutions matter, brand preferences matter, and missing one item can ruin the whole plan. If the AI regularly gets it wrong, the feature becomes a novelty. If it gets it mostly right—and learns quickly—it becomes a new default.

And that’s the quiet power move: Uber doesn’t need AI to be magical. It just needs it to make the cart feel less like work.