Google turns retail to agentic commerce

The National Retail Federation (NRF) is regarded as the central stage for the "future of retail" - with a focus on technology, customer experience, omnichannel and efficiency. In this setting, Alphabet and Google CEO Sundar Pichai clearly locates the current platform change as an AI platform shift.

The core of the Google thesis: Shopping is becoming "agentic"

Google's term for this is agentic commerce - AI agents perform tasks on behalf of customers, from product searches to transactions and aftercare. Google CEO Pichai describes a fundamental change: AI Mode shifts the search from keywords to natural conversations, with AI taking over the pre-sorting. Google cites the Shopping Graph with over 50 billion product listings as the database; more than 2 billion of these are updated every hour.

What Google is specifically announcing

1) The Universal Commerce Protocol (UCP): common language 

UCP is a new open standard that acts as a common language for AI agents, merchants and payment providers across the entire customer journey. UCP establishes a standard for agents and systems to collaborate across user interfaces, businesses and payment providers. UCP is designed to be cross-industry and compatible with existing protocols such as Agent2Agent (A2A), Agent Payments Protocol (AP2) and the Model Context Protocol (MCP). UCP was developed together with industry leaders such as Shopify, Etsy, Wayfair and Target and is supported by more than 20 other companies in the commerce ecosystem, including American Express, Stripe, Visa and Zalando.

2) Business Agent: brand chat directly in Google Search

The Business Agent is also new: customers can chat directly with brands in Search - like with a virtual sales consultant, in the "Brand Voice". Activation and customization take place via the Merchant Center; in the future, training with own data, insights, offers and direct purchases are planned.

3) Direct Offers: Offers in AI mode as a new ad play

This new pilot project for Google Ads enables advertisers to present exclusive offers, such as a special discount, directly in AI fashion to customers who are ready to buy. When users ask questions about products (e.g. looking for a carpet for a dining room), the AI can not only suggest suitable products, but also display exclusive offers directly. The display is AI-controlled: with Direct Offers, retailers specify in their campaign settings which offers they want to highlight and the AI decides when an offer is relevant.

4) Gemini Enterprise for Customer Experience: Shopping + Service as one agent platform

On the cloud side comes Gemini Enterprise for Customer Experience (CX): Google is introducing a central platform that combines shopping and customer service. Based on Google's latest Gemini models, ready-made and configurable AI agents are available there that can be implemented by companies within a few days. These agents can accompany the entire customer lifecycle, from the discovery of new products to solutions for potential problems after the purchase. The focus is shifting from simple chatbots to proactive digital concierges that can independently solve problems and carry out transactions under the supervision of companies. US retailers Lowe's and Kroger are among the first users of these agent tools, while Papa John's is the first company to use Google Cloud's omnichannel ordering agent (Food Ordering Agent).

What does this mean for the Swiss retail trade?

Google's NRF story can be boiled down to a simple key question: How do you not only get found in AI interfaces, but how do you get to the conclusion?

1) Product data becomes the "language" of AI

When shopping becomes conversational, classic keyword feeds are no longer enough. Product data must be able to answer questions: What goes with what, what alternative is available, which accessories are compatible - and why is the product the right choice. This is exactly where Google's Merchant Center logic is heading: away from a pure catalog and towards a data foundation that enables consultation in dialog.

2) "Buy where you browse" massively shortens the journey

With UCP and the checkout in AI Mode/Gemini, Google is moving the conclusion of the purchase directly to the moment of research. Shopping is thus completed more "in the conversation" - fewer click paths, fewer bounces, faster conversion. This is an opportunity for retailers, but also an architectural issue: as soon as the checkout takes place on platform interfaces, the customer relationship must still remain with the retailer.

This is why an early reality check is worthwhile for Swiss retailers: Are loyalty mechanisms (member prices, bonuses, benefits) already in place at the moment of purchase? Do transaction data and preferences flow cleanly back into your own CRM? And does the after-sales process - from service to re-engagement - remain in your own hands? In short: platform checkout can accelerate transactions. The decisive factor is that it also strengthens the relationship that counts afterwards.

3) Brand Voice becomes operational

With the Business Agent, Google puts the brand where retail is most sensitive: in the decision-making moment. Not as a banner, not as a claim, but as a conversation partner. This is precisely what makes the shift for branding so explosive: "brand voice" is transformed from a communication asset into an operative sales instance. Suddenly, tonality is no longer something that is polished in campaigns - it becomes a conversion lever in dialog. If you want to win here, you need more than a brand book in PDF format: You need an agent playbook that translates brand management into rules. In other words, in clean language, clear product truths instead of vague promises, defined no-goes and a brand safety that is capable of dialog and does not protect advertising environments, but answers.

What's more, goodwill is branding - and is "communicated" early on in agent-based shopping. Returns, exchanges, guarantees and delivery issues are not just service details, but part of the brand personality. And even cross-selling takes on a new quality: not as an upsell trick, but as helpful relationship management at the right moment, like a digital concierge. This is an opportunity for the Swiss retail trade because the sector traditionally thrives on advice, reliability and trust - and it is precisely these strengths that can now be scaled digitally. The key question is therefore not: "Do we need an agent?" But rather: If our brand speaks in the future - how should it sell?

Conclusion

Google is not presenting a "search refresh" at NRF 2026, but a new architecture: AI search as an assistance interface, UCP as a transaction layer, agents as new touchpoints - and ads as contextual completion assistants. The task for Swiss retailers is to prepare data, offers, brand voice and service in such a way that they function in an agent-based shopping world.