The news: Amazon is working to convince advertisers of the potential of AI shopping assistant Rufus. A leaked Amazon pitch deck cited by Adweek claims that Rufus is a “virtual product expert” and outlines how Rufus ads will be priced and measured as availability increases.

The pitch deck says that, as Rufus ads move from open beta to launch, advertisers will begin being charged and will be able to see cost-per-click data. Amazon is planning on a wide release “soon,” but did not specify when.

Catching up: Amazon has spent the past year testing ads in Rufus to foster an AI-driven retail media environment.

• Amazon began testing ways to embed sponsored product offerings into Rufus responses in 2025. This approach surfaces ads in AI-curated interactions, where Amazon’s algorithm determines which relevant products users see.
• The company has tested shoppable ads in Rufus that tap Amazon’s closed-loop ecosystem, allowing for a seamless path from product discovery to purchase.

Rufus has seen rapid adoption; Amazon noted 250 million active customers using the tool in 2025, and pointed out that users interacting with Rufus are 60% more likely to complete a purchase.

The trend: Retail media competitors are making similar moves. Walmart started testing ads within its AI shopping tool Sparky in 2025, including sponsored prompts that serve click-to-buy ads, and officially started running ads in Sparky earlier this year. Walmart is now introducing a shopping app within ChatGPT powered by Sparky, giving ChatGPT users the ability to check out via Walmart’s payment systems.

AI chatbots are pushing into commerce more broadly. Major platforms like ChatGPT and Microsoft Copilot embed ads directly into AI-powered conversations to reach users at high-intent moments; ChatGPT is also ramping up commerce features through partnerships with payment providers like PayPal.

Implications for marketers: Rufus offers an environment where a sponsored ad can link directly to a purchase—but marketers must know what makes consumers receptive to AI shopping assistants to succeed.

• About two-thirds of consumers have already shopped with a conversational AI assistant or are open to doing so, per 2025 Nosto data. Twenty-percent have used genAI tools like ChatGPT for shopping; 13% have interacted with AI assistants on brand websites.
• Despite the interest, trust is fragile. Sixty-nine percent of shoppers have abandoned their search because of irrelevant recommendations, indicating low tolerance for weak performance.
• Marketers should prioritize AI shopping assistants that align with consumer preferences and deliver useful, personalized guidance. The most appealing use cases include alerts for discounts and price drops, personalized recommendations, and gift inspiration, per Nosto, suggesting marketers should focus on assistants that help shoppers save time, discover relevant products, and act on purchase intent.