Google Just Made a Full-Stack Bet on Your Customer. Here’s What It Means.

From protocol layer to customer experience platform, Google is building the infrastructure to own the agentic commerce value chain. The strategic question isn’t whether to participate — it’s how much dependency you can afford.

When a technology company launches both an open protocol and a proprietary platform in the same week, that’s not a product announcement. That’s a strategic encirclement.

At NRF 2026, Google didn’t just introduce the Universal Commerce Protocol (UCP) — the open standard for agentic commerce I covered in my previous analysis. It simultaneously launched Gemini Enterprise for Customer Experience, an AI-native platform that handles shopping, customer service, and food ordering through autonomous agents — all powered by the same data that flows through UCP.

If UCP is the road, Gemini Enterprise for CX is Google’s vehicle on that road. And as Sundar Pichai told the NRF audience, Google intends to be both the infrastructure provider and a leading operator on top of it.

For C-suite leaders, this demands a more nuanced strategic calculus than simply “adopt the protocol.” The question is: where on the platform dependency spectrum does your brand want to live?

The Full-Stack Architecture

To understand Google’s play, you need to see the full stack:

Layer 1 — Data Foundation. Google’s Shopping Graph indexes 50 billion+ product listings with 2 billion refreshed per hour. This is the single largest structured commerce dataset on the planet. Every UCP transaction feeds it. Every Gemini CX interaction enriches it.

Layer 2 — Protocol (UCP). The open standard that allows any AI agent — Google’s, OpenAI’s, Microsoft’s, or an independent startup’s — to discover products, compare options, initiate checkout, and manage post-purchase. Open by design, with 20+ coalition partners including Walmart, Target, Best Buy, and Visa.

Layer 3 — Platform (Gemini Enterprise for CX). Three agent capabilities built on Google Cloud:

• Shopping Agent: Surfaces personalized product recommendations using the Shopping Graph, customer history, and real-time context. Already being tested by Walmart through its “Sparky” integration.
• CX Agent Studio: A no-code platform for retailers to build custom AI agents tailored to their brand voice, policies, and workflows. Lowe’s and Woolworths are early adopters.
• Food Ordering Agent: Handles the entire restaurant ordering flow — menu, customization, payment, delivery — autonomously. Papa John’s is also among the early adopters, according to a Google Cloud press release.

Layer 4 — The Flywheel. This is the strategic insight most analysis misses. UCP transactions generate commerce data that flows outward through the protocol to the ecosystem. Gemini CX interactions generate customer intelligence that flows inward to Google’s models. The more brands adopt UCP, the smarter Gemini CX becomes. The smarter Gemini CX becomes, the more value it creates for brands — which drives further UCP adoption.

This is a compounding data advantage. And it’s the real reason Sundar Pichai personally keynoted NRF.

The AWS Playbook, Applied to Commerce

If this architecture feels familiar, it should. Amazon pioneered this exact pattern with AWS: build the infrastructure that everyone uses, then build proprietary services on top of that infrastructure that create customer lock-in.

Google is applying the same logic to commerce:

• AWS ≈ UCP — Open infrastructure that becomes the industry standard. Let everyone build on it. Collect network effects.
• Amazon Marketplace ≈ Gemini Enterprise for CX — Proprietary services that deliver superior results because they sit on top of the infrastructure, with privileged access to the data flowing through it.

The difference? Amazon’s platform eventually competed with the merchants using it. Google insists UCP is open and that merchants remain the “merchant of record.” Whether that commitment holds at scale is the trillion-dollar governance question.

The Dependency Spectrum: A Framework for Executive Decision-Making

Every brand will land somewhere on this spectrum. The strategic art is choosing your position intentionally rather than drifting into one.

The Independent (Left of Spectrum)

Adopts UCP for interoperability but builds proprietary agent capabilities in-house. Invests in first-party data infrastructure, custom AI agents, and direct customer relationships. Highest cost, highest control, lowest platform risk.

Who this fits: Large retailers with engineering capacity (Walmart, Target), luxury brands with strong D2C relationships, brands with irreplaceable first-party data advantages.

The trade-off: You’re building what Google is offering for free (or near-free). But you own the customer relationship and the data it generates.

The Pragmatist (Center of Spectrum)

Adopts UCP + selectively uses Gemini CX components. Hedges by also supporting ACP and maintaining Amazon presence. Balances efficiency with control. This is where BCG recommends most mid-market retailers position.

Who this fits: Mid-market retailers, multi-brand portfolio companies, anyone who needs agentic commerce capability but can’t justify building from scratch.

The trade-off: You get Google’s scale and intelligence, but your customer data feeds their flywheel. You’re renting, not owning.

The Dependent (Right of Spectrum)

Goes all-in on Google’s stack — UCP, Gemini CX, Google Pay, Shopping Graph integration. Maximum efficiency, fastest time-to-market, but maximum platform dependency.

Who this fits: SMBs, emerging D2C brands, categories where speed-to-agent matters more than long-term platform strategy.

The trade-off: If Google’s priorities shift — and Scot Wingo’s Retailgentic analysis warns that platform priorities always shift eventually — you have limited optionality.

The Amazon Silence, Revisited

In my previous post, I flagged Amazon’s conspicuous absence from both UCP and ACP coalitions. That silence is even more significant viewed through the full-stack lens.

Amazon doesn’t need Google’s protocol or Google’s platform. It has its own Shopping Graph (the product catalog), its own agent infrastructure (Alexa+, Rufus), its own checkout and payments (Amazon Pay), and the deepest first-party purchase data in commerce history.

Amazon’s strategic posture suggests it views agentic commerce not as a protocol interoperability challenge, but as a walled-garden opportunity. A future where “Alexa, reorder everything I need for the week” doesn’t query UCP or ACP — it queries Amazon’s proprietary intelligence.

For brands, this means the three-body problem I identified earlier isn’t going away. Google (open protocol + platform), OpenAI/Stripe (conversational checkout + payments), and Amazon (closed ecosystem) represent three fundamentally different operating models. You need a strategy for each.

The Human-Centered Counterpoint

There’s a risk in this analysis of reducing everything to protocols and platforms. My colleague Justin Racine — Principal of Unified Commerce Strategy at Perficient and a four-time CMSWire Contributor of the Year — has been a voice for the human side of this equation.

In his NRF 2026 coverage, Justin makes a compelling case that while Google and OpenAI were announcing protocols, the broader NRF conversation was returning to fundamentals: trust, empathy, closeness. AI as an “intimacy engine” that deepens human connection, not a replacement for it.

His work on agentic and generative commerce in B2B extends this argument: even as AI agents automate transactions, B2B buyers still value — and will pay premium for — genuine human relationships.

This isn’t a contradiction of the platform thesis. It’s a constraint on it. The brands that win in agentic commerce won’t be the ones that simply plug into Google’s stack. They’ll be the ones that use the protocol efficiency to free up resources for the human moments that agents can’t replicate

Call to Action: Five Strategic Questions for Your Next Board Meeting

If agentic commerce isn’t already on your board agenda, it should be. Here are the five questions that will drive the conversation:

1. What percentage of our revenue is exposed to agent intermediation in the next 24 months? Bain estimates 15–25% of online retail by 2030, but some categories (commodity goods, replenishment, comparison shopping) will be disrupted much sooner. Map your category timeline.

2. Where do we want to sit on the platform dependency spectrum? This is a capital allocation decision, not an IT decision. Building proprietary agent capability costs 10–50x what adopting Gemini CX costs — but the long-term strategic implications are inversely proportional.

3. What is our first-party data strategy in an agentic world? If agent interactions flow through Google or OpenAI, who owns the customer intelligence? HBR’s research shows that only 6% of companies fully trust AI agents with core processes. Trust follows data ownership.

4. How do we measure success when the customer never visits our website? Traditional KPIs (site traffic, time on page, conversion rate) become meaningless in agentic commerce. You need new metrics: agent inclusion rate, agent recommendation rank, agentic conversion rate, and customer retention through agent channels.

5. Who owns this internally? Agentic commerce sits at the intersection of product data, digital commerce, marketing, and IT infrastructure. BCG’s 2026 Retail Executive Agenda recommends dedicated cross-functional ownership with direct C-suite reporting.

The Bottom Line

Google’s full-stack bet on agentic commerce — protocol + platform + data flywheel — is the most significant structural shift in digital commerce since mobile. It’s not a feature. It’s a new operating model.

The brands that move now — auditing data readiness, choosing their dependency position, investing in agentic capability — will have 18–24 months of advantage over those that wait. In a market McKinsey projects at $1–5 trillion by 2030, early positioning isn’t just valuable. It’s existential.

For the practitioner-level deep-dive on UCP architecture, functional primitives, and the Gemini CX technical stack, see my two-part Substack series: Part 1: The Protocol | Part 2: The Platform

Want to discuss agentic commerce strategy for your organization? Connect with me on LinkedIn.