Would you let robots spend your money?

Google's Universal Cart: AI Agents That Shop — and Spend — for You

Just days after Google I/O 2026 wrapped in Mountain View, the industry is still buzzing about the company's boldest step yet into AI-driven commerce. The new "Universal Cart" lets Gemini act as your personal shopper across unrelated retail sites, Google services, and , soon , YouTube and Gmail. The question on everyone's mind is simple: would you hand over your payment details to a robot?

At its core, the Universal Cart is an AI agent that can browse, compare prices, add items to a single checkout flow, and complete purchases without you leaving the Gemini interface. Google demonstrated the feature working seamlessly between its own shopping graph and third-party merchants. The system pulls real-time inventory, applies coupons automatically, and even suggests alternatives based on your past preferences. Integration with YouTube means product recommendations from review videos could turn into one-click carts inside the same app. Gmail could surface order confirmations or track shipments without ever opening a separate tab.

For consumers the appeal is obvious: frictionless shopping. Instead of juggling browser tabs or remembering which site had the best deal on headphones last week, you simply tell Gemini, "Find me the best noise-cancelling headphones under $200 and buy them." The agent handles the rest. Early testers at the I/O demo reported saving several minutes per transaction and discovering better prices they would have missed manually.

Yet the convenience comes with new risks. Giving an AI agent persistent access to payment methods and shipping addresses raises fresh questions about security, data privacy, and accountability. What happens if the model hallucinates a product detail or falls for a sophisticated phishing storefront? Google says it is building in human-in-the-loop confirmations for purchases above certain thresholds, but the exact guardrails are still being refined.

From my vantage point in Tokyo, these developments land differently than they do in Silicon Valley. Japan's e-commerce market is already highly mature, dominated by Rakuten, Amazon Japan, and Yahoo! Shopping. Consumers here prize reliability and after-sales service more than pure speed. A tool that can jump across merchants could pressure domestic platforms to open their APIs or risk losing visibility. At the same time, strict Japanese data-protection rules and consumer expectations around privacy mean Google will need to localize its consent flows carefully if it hopes to gain traction.

China's ecosystem offers another instructive contrast. Platforms such as Taobao and JD already embed powerful recommendation engines and one-tap payments, but they operate inside closed gardens. Google's cross-site approach could appeal to users frustrated by walled gardens, yet regulatory scrutiny on foreign AI services remains high. Southeast Asian markets, where mobile-first shopping is growing fastest, may prove the quickest testing ground for the Universal Cart outside the United States.

Beyond retail, the announcement signals a broader shift toward autonomous AI agents. Once Gemini can spend money responsibly, the same architecture could handle travel bookings, subscription management, or even small business procurement. That trajectory raises macroeconomic questions: how do we audit algorithmic purchasing decisions at scale? Will merchants need new APIs just to stay discoverable? And what new forms of digital fraud might emerge when the buyer is a language model rather than a human?

Google is not alone. OpenAI, Anthropic, and several Asian startups are racing to build similar agentic commerce layers. The company that first earns broad consumer trust , through transparent audit logs, clear refund policies, and ironclad security , will likely set the de-facto standard for the next decade of online shopping.

As someone who has covered Asia's tech scene for years, I see the Universal Cart less as a finished product and more as an early prototype of an entirely new interface layer between humans and markets. Whether it becomes as commonplace as search or remains a niche power-user feature depends on how thoughtfully Google addresses the trust gap it has just created.

This is Kenji Tanaka for Global1.news, reporting from Tokyo.

Source: The Verge via YouTube — 2026-05-19T18:41:47+00:00.