← 返回 Avalaches

该专栏认为,零售正在转向「agentic commerce」,也就是由自主式 AI 聊天机器人而非人类来挑选并购买商品;这可能像 10 多年前行动装置那样重塑数位购物,但前提是消费者愿意采用,且零售商能在不牺牲利润率的情况下参与其中。对 AI 辅助购物的兴趣在 9/2025 激增,原因是 OpenAI 在 ChatGPT 内推出「Instant Checkout」,让美国消费者不必离开对话,就能搜寻、选择、付款并安排出货。Walmart、Etsy、Shopify 等零售商已加入,Google 与 Microsoft 也在其 AI 工具中支援在聊天内购买。

零售商与平台正在加速行动:Wayfair 与 Google 合作推出 AI 驱动的购物;JD Sports 则与 Commercetools 达成协议,让美国客户可透过包括 Microsoft Copilot、Google Gemini 与 ChatGPT 在内的 AI 平台直接购买。为了让机器人能找到,品牌需要比传统 SEO 更明确、更结构化的产品资料,等同于一种「generative engine optimization」作法(例如,用具体成分,以及「无香味/添加剂/防腐剂」等属性来描述适合湿疹的护肤品)。机器人可能较少依赖星级评分等汇总指标,而更看重评论文字的实质内容;同时,广泛铺货、具竞争力的定价,以及避免缺货等经典因素,仍会影响商品是否被呈现并促成转换。

好处是透过个人化与探索带来更高销售,但风险集中在费用、控制权与资料:ChatGPT 透过 Instant Checkout 向商家收取未披露的每笔交易费用;Google 与 Copilot 目前不收取佣金,但平台日后可能会对搜寻版位、聊天内广告,或提供顾客旅程洞察收费。若更多营收转入 AI 生态系,零售商的广告业务与市集经济可能承压,因为购物者可能不再点进零售商网站,而 AI 平台掌握的行为资料可能比卖家从交易中获得的更有价值。文章指出谁是赢家仍不确定(并提到「social commerce」的热潮并未完全成真,即使 TikTok Shop 崛起而 Instagram 落后),同时对比 Amazon 以 Rufus 维持生态内控制,与 Walmart 的多轨策略(Sparky 加上与 OpenAI 在 10/2025 宣布、以及与 Google 在 1/2026 宣布的合作伙伴关系),并在结论主张零售商应立即行动而不是等待。

864f32bebbf3.png


81a5e0f6b8e1.png



The column argues retail is shifting toward “agentic commerce,” where autonomous AI chatbots, not humans, select and buy goods, potentially rewiring digital shopping the way mobile did more than 10 years ago, but only if consumers adopt it and retailers can participate without sacrificing margins. Interest in AI-assisted shopping spiked in 9/2025 after OpenAI unveiled “Instant Checkout” inside ChatGPT, letting US consumers search, choose, pay, and arrange shipping without leaving the chat. Retailers such as Walmart, Etsy, and Shopify have signed up, and Google and Microsoft also enable in-chat purchasing within their AI tools.

Retailers and platforms are moving quickly: Wayfair has partnered with Google on AI-powered shopping, and JD Sports struck a deal with Commercetools to let US customers buy directly through AI platforms including Microsoft Copilot, Google Gemini, and ChatGPT. To be discoverable to bots, brands need more explicit, structured product data than traditional SEO, effectively a “generative engine optimization” approach (for example, describing eczema-suitable skincare with concrete ingredient and “no scent/additives/preservatives” attributes). Bots may rely less on aggregated metrics like star ratings and more on the text substance of reviews, while classic factors like broad distribution, competitive pricing, and avoiding out-of-stocks still influence whether products surface and convert.

The upside is higher sales via personalization and discovery, but the risks concentrate in fees, control, and data: ChatGPT charges merchants an undisclosed per-transaction fee via Instant Checkout, while Google and Copilot are not currently charging commissions, yet platforms could later charge for search placement, in-chat ads, or insight into the customer journey. If more revenue shifts into AI ecosystems, retailer ad businesses and marketplace economics could be pressured because shoppers may not click through to retailer sites, and AI platforms may own more valuable behavioral data than the seller receives from the transaction. The piece notes uncertainty about winners (citing how “social commerce” hype did not fully materialize, even as TikTok Shop rose while Instagram lagged), contrasting Amazon’s in-ecosystem control with Rufus versus Walmart’s multi-track approach (Sparky plus partnerships with OpenAI announced 10/2025 and Google announced 1/2026), and concludes retailers should act now rather than wait.
2026-02-19 (Thursday) · 70adcb795ebfc9d9b4d4143fa06f252162905ec4