Google Launches Universal AI Shopping Cart Ahead of the Holidays

Google Launches a Universal AI Shopping Cart

Google has begun rolling out Universal Cart in the US — an AI-powered tool that lets shoppers save products from multiple retailers into a single, unified cart accessible through Google Search and the Gemini app. First previewed at Google I/O in May, the tool is launching just in time for Black Friday and the holiday shopping season.

Here's how it works: find a product via Google Search or Gemini, save it to your universal cart, and come back later — your items are still there, even if they're spread across several different stores. The system also sends notifications about price drops or restocked items, and can surface relevant promo codes. Paired with Google Wallet, it can analyze the perks of a user's different credit cards and recommend the best payment method for each purchase.

Under the hood, checkout relies on Google's Universal Commerce Protocol (UCP) — an open standard the company introduced earlier this year to let AI agents and retailers cooperate throughout the shopping journey, including at checkout.

The launch comes as competition in AI-assisted shopping heats up: Amazon, OpenAI, and Meta have all rolled out similar agentic shopping tools, and Shopify recently opened its massive catalog (over a billion items) to more AI shopping agents. For small and medium e-commerce businesses, the takeaway is clear: more shoppers will discover your products not just through traditional search, but through AI agents that compare, save, and decide on their behalf — making it worth double-checking that your catalog is clean and machine-readable.

Sources: The Verge | TechCrunch

A Practical Warning: AI Agent Costs Are Set to Jump

Separately, research firm Gartner published a report warning of what it calls the "Inference Paradox": even though per-token prices keep falling, the total cost of running AI agents in the enterprise is projected to rise more than fivefold by 2028. The reason: AI agents consume 5 to 30 times more tokens than a basic chatbot task, and AI labs are reinvesting token-price savings into newer, pricier frontier models — so the savings rarely reach the end customer.

Gartner recommends five practical steps for businesses: invest in token-usage observability (FinOps), tie every AI deployment to a measurable business outcome, automatically route simple tasks to cheaper models, set minimum quality/compliance thresholds before deploying any new agent capability, and use proprietary company data to build smaller, more efficient specialized models instead of relying solely on generic frontier models.

It's a useful reminder for small and medium businesses considering AI agent adoption: the excitement is real, but upfront budget planning matters just as much.

Source: VentureBeat