When a shopper browses a storefront or a seller tweaks a price and uploads photos, everything looks business as usual on the surface. Behind the scenes at major global platforms, though, built-in AI now handles all three layers: one tool builds product backgrounds, another drafts store discounts, and a third decides which item appears in a chat answer. While these Western tools don't operate directly in Russia, they clearly signal where the market is headed and what rules already govern international storefronts.
Google Product Studio in Merchant Center: Scene Generation, 70-Image Cap, and Mandatory IPTC Tags
Start with the storefront. Google Merchant Center is Google's seller hub where merchants upload product feeds—prices, stock, and photos—for Search and Google Shopping. Previously, getting lifestyle backgrounds meant booking a studio or tweaking images in external generators and resizing them manually. Google Product Studio builds that step right into the dashboard: upload a source photo, remove the background, upscale resolution, or place the item in a new scene. Describe the subject, placement, surroundings, and backdrop, review four generated options, and save the winner straight to your listing.
In practice, strict guardrails apply. Product Studio isn't available in Russia and works only in supported countries. Each item is capped at 70 images—once slots are full, new photos won't save until you delete old ones. Regulated categories—alcohol, tobacco, pharmaceuticals and medical devices, weapons, fireworks, and gambling—are blocked. Complex visual elements like human hands, wall art, and lighting fixtures aren't supported either.
There is also a hidden rule inside every file. Google embeds an IPTC provenance tag (DigitalSourceType set to TrainedAlgorithmicMedia) into each generated image—a digital metadata passport marking the frame as AI-created.
Google strictly prohibits stripping IPTC tags from generated files. If an aggressive image optimizer or CDN strips metadata on delivery, the image violates policy. Before uploading your feed, verify that your pipeline preserves metadata fields.
Square Squaremates: An Operations AI Assistant That Requires Manual Approval
While Google tackles visuals, US commerce platform Square went after the back office. Many sellers expect an AI 'magic button' that reprices items and launches promos on its own. Square built the opposite: the assistant prepares routine work, but cannot hit 'Apply' without your click.
Called Squaremates, the tool was previously tested in open beta as Managerbot and now serves as the main entry point in Square Dashboard. It answers questions about sales, orders, inventory, and customers using live account data, and prepares drafts for catalog edits, discounts, customer emails, and staff schedules.
Every action runs only after the seller explicitly approves it. Square itself warns that responses may be inaccurate—one bot mistake on a discount could wipe out margin in hours, so manual sign-off protects working capital.
Access has clear boundaries too. Squaremates works in English for single-account US sellers—Multi-Account Suite isn't supported yet. The assistant itself (along with developer tools Square MCP and Square CLI) comes at no extra charge, though advanced actions like expanded marketing or scheduling still require a paid Square plan.
Walmart Sparky Grows 129% Year-Over-Year: Why Clean Product Attributes Matter More Than Ever
The third loop is the shopper. Inside Walmart, AI assistant Sparky helps customers search and compare products through conversation instead of clicking through filters. On November 19, 2026, during Walmart's Q3 FY27 earnings call, President and CEO John Furner shared fresh numbers: total revenue rose 5.2% to $188.8B, global e-commerce grew 17%, and US usage of Sparky jumped 129% year-over-year.
Another notable figure stood out: customers building orders with Sparky generate average baskets roughly 40% larger than non-Sparky shoppers. Keep a level head here, though. Walmart didn't disclose total active users or the share of chats that convert into completed purchases. Higher usage and bigger baskets don't prove every individual listing automatically converts better.
What is undeniable is the rising bar for data hygiene. When shoppers ask an AI assistant for recommendations, it pulls facts from product attributes—structured fields covering size, material, compatibility, variants, and intended use. If those fields are sloppy or stuffed with legacy search keywords, the assistant won't understand who the product fits and will recommend a rival listing instead. On global storefronts, clean attribute data already decides whether algorithms surface your product.

