The Cart That Never Sleeps
A shopper in Denver adds a $349 espresso machine to her cart on a Tuesday night. Not a retailer's cart. Google's cart. She closes the tab and forgets about it.
The cart does not forget. It watches the price while she sleeps. It knows the machine sold for $289 in March, because it keeps the chart. When the price dips, when stock runs out and returns, when a better offer appears somewhere else, it tells her.
She never installed a price tracker. She just shopped the way she always does.
At I/O on May 19, Google launched Universal Cart, a single cart that spans merchants and follows shoppers across Search, the Gemini app, YouTube, and Gmail. Google's announcement describes a cart that works in the background: it finds deals and price drops, surfaces price history, and sends alerts when items return to stock. It runs on Gemini models, and it rolls out in the US this summer, starting with Search and the Gemini app.
TechCrunch described it as a cart that wants to "follow your entire shopping journey across the internet."
Most coverage filed this under convenience. That reading is too small. Google just deputized every shopper as a price-tracking bot. The consequences land on pricing teams, not consumers.
Price Tracking Used to Be a Power Tool
For fifteen years, price tracking was a niche habit. CamelCamelCamel charted Amazon prices for deal hunters. Keepa served resellers. Honey hunted coupon codes at checkout. The tools worked, but they demanded setup, and setup filters out almost everyone.
Universal Cart removes the setup. Adding an item to a cart is the lowest-effort action in commerce. People do it idly, the way they bookmark articles they never read. Under Google's model, that idle gesture enrolls an agent that monitors the item indefinitely: its price, its history, its availability, its alternatives.
The scale is the story. Google says people shop across its surfaces more than a billion times a day, drawing on a Shopping Graph that now holds over 60 billion product listings. Every one of those sessions can leave a watcher behind.
That is what ambient means. Nobody decides to track prices anymore. Tracking is simply what carts do. And when monitoring becomes the default, the monitored population stops being deal hunters and becomes everyone.
What the Cart Sees That Your Team Does Not
Here is the uncomfortable asymmetry. A shopper's free cart now runs sharper price intelligence than many brands' paid tooling.
| Signal | Universal Cart, shopper side | Typical brand practice |
|---|---|---|
| Price drops | Watched continuously, alerts pushed automatically | Weekly spot checks in a spreadsheet |
| Price history | Charted per item, shown at the moment of decision | Tracked for own SKUs, rarely for rivals |
| Restocks | Cross-merchant alerts the moment inventory returns | Own warehouse feeds only |
| Cross-merchant comparison | The default view, because the cart spans retailers | Quarterly pricing studies |
| Buy timing | Agent waits for a threshold, then acts | Promo calendar locked weeks ahead |
Three blind spots bite hardest.
- Price history is public memory now. Your January clearance shapes what a May shopper believes the product is worth, because the cart shows them the chart. Anchor pricing assumes buyers forget. Agents do not.
- Stockouts become transfers. When your listing goes dark, restock alerts route the waiting demand to whichever merchant fills the gap first. An out-of-stock week is no longer paused revenue. It is donated revenue.
- Your compare set is whatever the model says it is. Google demoed Gemini flagging incompatible PC parts inside a mixed cart and proposing alternatives. The same logic applies to substitutes everywhere. You compete with everything the model considers a credible swap, whether or not it sits on your tracking list.
The Merchant Tools Confirm the Shift
One day after I/O, Google Marketing Live handed merchants their side of the bargain. Buy for Me lets Google complete a purchase on a merchant's own site through Google Pay. Launch partners include Nike, Sephora, Target, Walmart, Ulta Beauty, Wayfair, and Shopify merchants such as Fenty Beauty and Steve Madden. Checkout expands to Canada and Australia in the coming months, then the UK, with hotel booking and food delivery on the roadmap. Affirm and Klarna are being folded into Google Pay as well.
The quieter announcements matter more for data teams.
- Conversational Attributes. A new Merchant Center schema that lets retailers rewrite product data for conversational queries. Shoppers no longer type "stroller lightweight aluminum." They ask for one they can fold one-handed while holding a baby. Your feed has to answer that sentence.
- Ask Advisor. An AI collaborator inside Merchant Center that answers questions about your catalog and performance, connected to Google Ads and Analytics.
- AI performance insights. Competitive reporting that shows your share of voice against similar brands inside AI shopping surfaces, launching in the US, India, Canada, Australia, and New Zealand.
Read those together and the message is blunt. The agent's view of your catalog is now your storefront.
Then add AP2, Google's agent payments layer, which lets users cap what an agent may spend by brand and by budget. That sounds like a safety feature, and it is. It is also a pricing mechanism. A hard spending cap plus a continuous price watch produces an agent that buys the instant your price crosses a threshold you cannot see.
June 23 Will Be the First Agentic Prime Day
Amazon, meanwhile, confirmed that Prime Day 2026 runs June 23 to 26: four days, deals across more than 35 categories, and its first June event at that length. The date puts a deadline on everything above.
Consider the baseline. Adobe Analytics measured $24.1 billion in US online spend during last July's four-day event, up 30.3% year over year. Adobe also found that traffic to US retail sites from generative AI sources rose 3,300% year over year during that window. All of that happened before any platform shipped an always-on consumer cart.
Google's rollout may still be in progress when the event starts. The behavior it formalizes is already here. Digital Commerce 360 reported on May 22 that Walmart shoppers who use Sparky, the retailer's AI agent, carry an average order value roughly 35% higher than non-users. Weekly active users more than doubled over the past quarter, and units purchased through Sparky more than quadrupled. Walmart's CEO told investors the company is becoming AI native.
Agent-assisted shoppers do not browse. They execute.
So expect the first Prime Day where deals get arbitraged in minutes. A lightning deal that drops at 2 a.m. no longer waits for humans to wake up. Agents holding that SKU check it against their caps and move. If a rival undercuts you on day one, the carts holding both options will have decided long before your Tuesday pricing meeting convenes.
Match the Cadence or Fly Blind
You cannot out-scale a billion shopping sessions. You can match their cadence on the catalog that matters: your products, and the substitutes agents place beside them.
- Monitor continuously, not weekly. An hourly watcher takes 168 snapshots of your market position for every one a weekly review takes. Real-time price tracking across marketplaces, the kind BrandBaazar's pricing intelligence products are built for, exists to close that exact gap.
- Audit what agents see. Pull your listings the way an agent would: feed data, conversational attributes, availability, and the price history chart a shopper gets shown. If that chart contradicts your premium positioning, you want to know before June 23, not after.
- Treat restocks as market events. A competitor's restock now fires alerts to a queue of waiting buyers. A competitor's stockout opens a window measured in hours. You only catch either with live availability data.
- Set thresholds like an agent. AP2 budgets tell you how agents behave: they hold a line and act when it breaks. Define your own lines: the price gap, the share-of-voice dip, the stockout duration that demands a response. Then pipe alerts into your systems through an API instead of a Monday dashboard.
The Watchers Will Multiply
Universal Cart reaches YouTube and Gmail next, then more countries, then hotels and food delivery. Every expansion adds watchers. Amazon's four-day June event will hand them their first real stress test, and the holiday season will hand them their second.
Price used to be something you published. It is becoming something agents consume, compare, and act on around the clock, whether or not a human is looking at a screen.
Price intelligence used to be a product brands bought. As of May 19, it is a feature shoppers get free with their cart.
Your customers stopped flying blind this week. Make sure they are not the only ones.