Before we get started… This article was not written by AI. I’ve personally written this article, warts and all, based on conversations with my former colleagues at Google Cloud, public press releases, pricing pages, deprecated product docs, and a few backchannels and nuggets buried in TOS pages that I shan’t publicly disclose.
Who are we and why do we care? Nimstrata is a Google Cloud partner and we’re the only partners who have built a native Shopify App that allows retailers to connect their online stores directly to Google Cloud’s AI solutions. We see conversational commerce as the next evolution in online shopping and we’re helping Google deliver it.
What is Conversational Commerce?
Is it showing up in ChatGPT when someone asks about your category? Is it offering a chat-based interface on your website? Is it a series of quizzes and if/then logic to get people to the right products? The answer is yes. But just because conversational commerce can be all of these things doesn’t mean that it needs to be.
You should be focused on adopting conversational elements that actually matter for your niche and enrich the experience, not distract people.
Shopping Agent Announced
Google Cloud announced the Shopping Agent at NRF 2026 as part of their Gemini Enterprise for Customer Experience (GECX) release to stake their claim in the Conversational Commerce conversation. That's a mouthful... Sundar flew out to NRF for the event, so it shows that Google is serious about this.
Google’s argument is that having a unified platform allows brands to access all of Google’s knowledge of the world and not just use OpenAI or Anthropic to create ChatGPT or Claude wrappers. Google’s argument is, “Get all your data into Google and you’re golden.” If this sounds monopolistic, you’re probably right, but I tend to agree with them. Why? Because I’ve personally seen the results that Google’s AI deployed across Shopify storefronts can deliver.
So, what is the Shopping Agent?
It’s simply a “Pre-Built Agent” from Google Cloud with a basic set of instructions and rules to help customers discover and buy products on an ecommerce website. It contains system tools that can be used by the agent to perform common tasks. They are not editable.
In AI terms, agents have skills that call tools. Let’s break that down further:
Agent: An autonomous AI designed to complete tasks on a user's behalf. This is where “reasoning” happens. For example, “Go find me a red t-shirt.”
Skill: A reusable workflow consisting of instructions (or scripts) that enable AI agents to complete tasks consistently. For example, a skill may be searching and filtering within a product catalog.
Tool: A third-party API or endpoint that allows the agent to interact with the external world - just think of this as the interface or front door to other services. For example, connecting to Shopify’s catalog or cart.
Shopping Agent skills and tools are built, configured, and deployed inside of CX Agent Studio, more below.
What does it look like?
You can deploy the agent in three ways:
Create a Web Widget – This is probably the most common way and allows you to drop an agent on a website just like any another pre-AI-taking-over-the-world chatbot. You can vibe code this.
Connect to an Existing Platform – This is how you’d connect this logic and knowledge to Twilio, Google Telephony Platform, etc.
Set up API Access – This gives you the most control and flexibility over the agent’s look, feel, and on-site behavior. You can probably vibe code most of this, but don’t forget about the serving layer. You’ll need infrastructure somewhere to authenticate against these APIs, especially if you’re using a Liquid template (Theme) on Shopify – or use our storefront service.
How much does it cost?
Public pricing is currently $0.50 USD per interaction. That adds up fast at scale, so it's worth modeling against your traffic before committing.
CX Agent Studio
CX Agent Studio is a control layer of Gemini Enterprise for Customer Experience. The CX Agent Studio dashboard is not accessed via the Google Cloud Console, but this link instead: https://ces.cloud.google.com/. To use CX Agent Studio, the Customer Engagement Suite API will need to be enabled on the underlying Google Cloud project, which may require some help from IT or security if you’re trying this at work.
Note that this is different than Google’s (now legacy) conversational agents including Dialogflow CX and Dialogflow ES and their related playbooks. If you’re using an older Dialogflow solution, I’m sorry. It really sucks to be rug-pulled by Google on another product deprecation, but you probably already knew this was a possibility before climbing into bed with them.
Google’s Shopping Agent is just one piece of Gemini Enterprise for Customer Experience (GECX) and a great next step is evaluate it for customer support use cases as well.
Commerce Search Integration
The Shopping Agent is technically a completely separate service and offering from Google Cloud. However, the Shopping Agent has a tool that is powered by Commerce Search aka Vertex AI Search for Commerce. This means that the Shopping Agent leverages Google's Retail API when possible to deliver the best, most relevant and revenue-optimized results.
Unfortunately for early adopters of the Commerce Search (aka Vertex AI Search for Commerce) Conversational Agent, it's time to re-build. While the API will continue to work, it's effectively dead in terms of new features.
Recommendations AI continues to live on, owned by the same product team, but we'll keep watching this space because it's always kind of been the redheaded stepchild of the Retail API.
Universal Commerce Protocol (UCP)
In January 2026, alongside the Shopping Agent announcement, Google also introduced the Universal Commerce Protocol (UCP) which is an open standard co-developed with Shopify, Etsy, Wayfair, Target, and Walmart. Think of UCP as the plumbing that makes agentic commerce actually work across the internet.
The idea is supposed to be simple: instead of every AI agent building custom integrations with every retailer, UCP provides a common language for product discovery, checkout, and post-purchase workflows. Any AI agent, whether it's Google's AI Mode in Search, Gemini, or a third-party surface, can use UCP to browse your catalog, add items to a cart, and complete a purchase without the customer ever leaving the conversation.
A few things worth knowing:
You stay the Merchant of Record. Your customer data, your relationships, your post-purchase experience. Google doesn't insert itself into the transaction.
It uses your existing Merchant Center feeds. If you're already running Google Shopping, you're halfway there.
It's open source (Apache 2). This isn't a walled garden — it's designed to interoperate with A2A, MCP, and the Agent Payments Protocol (AP2).
It's early. Multi-item carts, account linking for loyalty programs, and returns handling are still on the roadmap.
For Shopify retailers, UCP means your store can be "agent-ready" — discoverable and purchasable by AI agents across Google's surfaces. This is where the Shopping Agent and Commerce Search converge: the agent reasons, Commerce Search finds the products, and UCP handles the transaction.
What’s next?
We think you schedule some time with us if you’re on Shopify and want to set up agentic or conversational commerce in addition to a better on-site search and merchandising experience. We’re undeniably the experts here, and we’re unapologetic about it because we’ve seen a lot of confusion, FUD, and misinformation published by other Google Cloud partners and Shopify agencies that we're brought in clean up. If you want it built right the first time, let us take care of it for you.
TL;DR
Shopping Agent is Google Cloud's new pre-built conversational commerce agent
It can be powered by Commerce Search under the hood to provide best-in-class search results
If you're on Shopify or Salesforce Agentforce Commerce, Nimstrata's Retail Cloud Connect apps connect you to all of this seamlessly
Read more articles
