The growing role of AI in shopping is forcing retailers to rethink discovery and decision-making.

Walmart, for example, is embedding its ecommerce capabilities into external AI assistants like Google’s Gemini. Meanwhile, Amazon is keeping AI-led discovery and decision-making inside its own ecosystem with tools like Rufus.

Here’s how their strategies diverge and why most retailers will land somewhere in between.

Walmart’s strategy: Be where the agents are

Walmart is adopting an “everywhere” approach to agentic commerce.

• Sparky, Walmart’s in-app AI assistant, helps shoppers compare products and synthesize reviews.
• Beyond its own app, Walmart is extending that functionality through a partnership with Google’s Gemini, designed to surface Walmart products directly within AI conversations and enable a seamless path to checkout.
• The company is also backing open commerce protocols to make retailers interoperable with multiple AI assistants.

By using external agents like Gemini and supporting open commerce protocols, Walmart is betting that volume will come from being the easiest retailer for agents to transact with.

• This gives Walmart access to more shopping moments, even outside its own app.
• However, if the AI becomes the main contact, Walmart becomes infrastructure, not the brand.

Amazon’s strategy: Make Amazon the agent

Amazon is taking the opposite tack, keeping the agent inside the ecosystem.

• Its AI assistant, Rufus, is embedded across Amazon’s app and site, guiding discovery and decision-making without sending users elsewhere.
• Amazon has said shoppers who use Rufus are 60% more likely to make a purchase, with the assistant on track to deliver over $10 billion in incremental sales annually.

Keeping Rufus inside Amazon’s ecosystem lets Amazon defend its most valuable assets: Search behavior, Prime loyalty, and retail media revenue.

• To Amazon, AI is not a threat to the existing flywheel, but rather accelerates it.
• But this strategy assumes consumers will tolerate a walled-garden agent, even as cross-retailer agents become more capable.

The workable strategy: Optimize for agent access, not agent ownership

Most retailers don’t have the resources to build out agentic systems like Walmart and Amazon. The viable path is becoming agent-ready.

That means treating AI agents as a new distribution layer, like search engines or marketplaces before them.

• Retailers will gain visibility by focusing less on flashy AI experiences and more on the fundamentals: Clean product data, transparent pricing and availability, fast fulfillment, and APIs that make inventory and checkout easy to surface wherever discovery happens.
• Retailers that are easy for agents to understand and transact with will appear more often in AI-led recommendations, regardless of the assistant the consumer uses.

The trade-off is that AI leaves less room for brand storytelling.

• When an agent decides what’s “best,” it often comes down to price, speed, and ratings.
• For mid-tier and specialty retailers, the advantage will come from offering something an AI cannot easily replace, whether that is unique products, exclusivity, loyalty benefits, or credibility built through expertise.

This was originally featured in the Retail Daily newsletter. For more retail insights, statistics, and trends, subscribe here.