Google is promoting a new wave of “agentic commerce,” where AI agents not only answer questions but help shoppers to complete tasks and transactions end to end. Google’s new Universal Commerce Protocol (UCP) and related tools are being adopted by major retailers like The Home Depot and Lowe’s to enable more seamless, personalized, and automated shopping experiences.
UCP is an open standard meant to give AI agents a common “language” to work with retailers, payment providers, and different consumer interfaces across the full shopping journey—from discovery to post‑purchase support. Instead of every retailer building one‑off integrations for each AI agent, UCP allows agents to interact in a unified way across surfaces like Search, Gemini, and retailer systems. Co‑developed with major players such as Wayfair, Target, and Walmart, and endorsed by Best Buy and The Home Depot, UCP reflects a broad industry attempt to standardize how agentic commerce works. One early visible feature is a new checkout option on eligible Google product listings in AI Mode in Search and in the Gemini app, where shoppers can buy directly during their search experience using Google Pay (with PayPal support to come), while retailers remain the seller of record and control how the integration is customized. Over time, Google plans to layer in richer capabilities like discovering related products and applying loyalty rewards directly within these AI‑driven flows.
The Home Depot is using Google Cloud’s AI to expand its Magic Apron assistant into an “AI‑first” experience that is more conversational, contextual, and omnichannel. Magic Apron is evolving from basic search to a project advisor that can guide customers through tasks ranging from fixing a leaky faucet to planning a full kitchen remodel, using personalized recommendations and how‑to support. By integrating Google’s Gemini Enterprise for Customer Experience Shopping agent, Home Depot is adding multimodal features such as image upload and visualization so the assistant can understand pictures and help with complex home improvement scenarios. In‑store, Magic Apron now connects to real‑time local inventory and precise product location, giving aisle‑level directions and technical guidance, so a shopper asking about grout for glass tiles can receive how‑to content, an appropriate product recommendation (such as unsanded or epoxy grout), and exact shelf location in one flow. The new in‑store capabilities are being piloted in select locations with plans for nationwide rollout later in the year. For pros, Home Depot has launched an AI‑powered materials list aggregator on its pro digital site: pros can describe a project via voice or text or upload an existing list, and the agent generates a comprehensive, grouped materials list—rolled out in beta in late 2025 and scaling nationally now.
Home Depot is also applying AI behind the scenes to logistics and corporate workflows. Its route precision system blends customer‑specific data (hours, drop‑off preferences) with external signals (weather, road quality) and multimodal analysis of Google Maps to spot barriers like narrow streets, gates, or posts that might complicate delivery, with the goal of predicting and preventing failures. At the Store Support Center, Gemini Enterprise powers “specialized agents” that automate internal tasks such as identifying project bottlenecks, generating marketing copy, and auditing digital designs, effectively bringing the same agentic approach inside the organization.
Lowe’s is another early adopter of Google’s new agentic tools, especially Business Agent, which lets brands “chat” with customers directly from search results and supports training agents on retailer data, offering related products, and enabling direct agent‑led checkout. Lowe’s is using Gemini Enterprise for Customer Experience to enhance its Mylow AI home improvement advisor, aiming to make projects less intimidating by tailoring guidance to a customer’s home, project, and location. The retailer plans to leverage Google’s Shopping agent, which can interpret images, videos, and voice inputs without typed queries, to do things like read a handwritten materials list, determine what products are needed, fill a digital cart, and apply member discounts. Google emphasizes that the Shopping agent acts only with explicit consent: it can factor in past local purchase history and real‑time availability to suggest options, and—with user approval—add items to carts and complete checkout. A visual drag‑and‑drop canvas is intended to let employees with any skill level design and launch support workflows in days, enabling sophisticated AI deployments with minimal engineering effort.
More information is available here.
Image above: Courtesy of Google for Developers.
You must be logged in to post a comment.