Amazon launches Alexa for Shopping, replacing Rufus

The most strategically aggressive piece of Alexa for Shopping is not the conversational search bar — it is the quiet expansion of 'Shop Direct' and 'Buy for Me,' features that let Amazon's AI browse other retailers' websites and place orders on a customer's behalf ^[1][2]. In effect, Amazon is positioning itself as the agentic front-end for shopping on the broader web, even when the inventory, fulfillment, and brand belong to someone else. The mechanics matter: orders arrive at third-party retailers from 'buyforme.amazon' email addresses, and Shop Direct's product coverage exploded from 65,000 SKUs at launch to more than 500,000 by November 2025, often without explicit merchant opt-in ^[7]. Some retailers say they have received orders for items they don't even sell, after Amazon scraped their catalogs ^[7]. The strategic logic is straightforward — own the user relationship, the discovery surface, and the checkout, regardless of which logo sits on the destination site — but it stretches the boundaries of consent, data ownership, and antitrust law, and it does so on top of Amazon's already-pending Perplexity litigation over agent access ^[6]. If Buy for Me sticks, Amazon doesn't just defend its marketplace; it taxes everyone else's.

Amazon's advertising business is roughly $56 billion, and most of it lives on the same search and product pages that Alexa for Shopping is about to colonize with conversational answers, AI overviews, and side-by-side comparisons ^[1][6]. That real estate has historically been the most expensive in e-commerce — third-party sellers pay top dollar for sponsored placements that surface their listings above organic results. When the assistant injects a curated recommendation set or comparison grid at the top of the page, those paid placements get pushed down or reinterpreted by the model, and merchants lose deterministic control over how their inventory shows up ^[4]. The trade-off Amazon is making is visible in the numbers: it has internal evidence that AI-assisted sessions convert at more than 3x the rate of non-AI sessions, and that Rufus alone is generating an estimated ~$12 billion in incremental annual sales ^[5]. If Alexa for Shopping scales that conversion lift across the entire shopper base, the GMV upside likely outweighs any near-term ad disruption — but in the medium term, expect Amazon to rebuild its ad products around AI surfaces (sponsored answers, sponsored comparisons, paid 'recommended by Alexa' slots), and expect 3P sellers to spend the rest of 2026 figuring out how to rank inside an agent rather than inside a search results page ^[4].

Strip away the launch theater and Alexa for Shopping is a bet that closed, retailer-owned agents will beat open, third-party agents on the two metrics that actually matter — trust and conversion. Bain's research finds that consumers trust retailers' on-site AI agents about 3x more than third-party agents, a structural advantage that's hard for ChatGPT, Gemini, or Perplexity to engineer their way around ^[5]. The infrastructure gap is just as stark: Forrester argues that OpenAI's Instant Checkout stumbled because of 'the absence of inventory management infrastructure…disastrously absent from the plan,' precisely the layer Amazon has spent twenty-five years building ^[5]. A Walmart executive has noted that purchase conversion inside ChatGPT runs at roughly one-third the rate of Walmart.com, reinforcing how thin the open-agent value proposition still is ^[5]. OpenAI has effectively wound down Instant Checkout, with only ~30 merchants live six months in, and is pivoting toward dedicated retailer apps inside ChatGPT ^[4][5]. Google's Gemini-based 'Buy for Me' and the A2A agent-to-agent protocol push the opposite, open-ecosystem vision ^[6]. Amazon's wager is that customers will pick the agent that already knows their address, their cards, their return history, and their Prime perks — and that the agent's owner controls the inventory. If that wager holds, agentic commerce ends up looking less like an open browser and more like a walled garden with a chat interface.

The reception has been notably cooler than the launch deck suggests. Inside Amazon's employee community, the rollout is being read primarily as a brand consolidation rather than a re-org — Rufus's product expertise and shopping history continue to power Alexa for Shopping under the hood ^[1]. Externally, the consumer reaction skews tepid: shoppers point out that core Alexa functionality (alarms, smart-home routines) still misfires and ask why Amazon is layering on a new agentic shopping surface before fixing the basics. Skeptics frame the move as a name change rather than a step-function product improvement. The optimist's counter is that brand matters more than novelty in this category. Gartner's Brad Jashinsky argues that 'the combination of putting Alexa for shopping front and center along with Alexa's greater brand awareness will lead to more usage' ^[4], and the numbers behind Rufus's quiet ramp — over 300 million customers served in 2025, MAUs up more than 115% and engagement up nearly 400% year-over-year — suggest that distribution, not novelty, is the real unlock ^[3]. The fairer read is that both camps are partly right: Alexa for Shopping is mostly a packaging and distribution move on top of a product (Rufus) that was already working, dressed up with a handful of genuinely new agentic features like Scheduled Actions and Buy for Me ^[2].