A shopping chatbot with 300 million users got sunset on May 13. In its place: Alexa for Shopping — a single agent that compares products, tracks prices, and completes purchases on a shopper's behalf, including on sites Amazon doesn't own. Agentic commerce stopped being a 2027 keynote slide this week and became default Amazon UX. The discovery layer marketers have spent 20 years optimizing is being rewritten in May.
What happened. On May 13, Amazon retired Rufus — the standalone chatbot that hit 300 million users by the end of 2025 — and folded it into Alexa for Shopping, one agent living across the app, the website, and Echo Show. The headline isn't recommendation. It's purchase: the new Alexa compares products, tracks prices, and completes buys on a shopper's behalf — including on third-party sites Amazon doesn't own.
Why it matters. Two years of "the agent is the new interface" talk collapsed into one launch from the largest retail-media network on earth. The 2026 shopper asks Alexa, "best quiet outdoor fan under $200, no plastic blades" — and the answer is a product, not a results page. Your title tags and bullets are now LLM training data. The agent also walks off Amazon when something fits better — so one buyer decides across both your Amazon page and your own site, and data cleanliness beats ad spend. Same architecture Walmart's Sparky and Alibaba's Qwen shipped this quarter. The agent layer went from experiment to default UX in 12 months.
HubSpot launched AEO Sensor, a free public dashboard tracking how ChatGPT, Gemini, and Perplexity cite brands by industry. It landed alongside HubSpot's finding that ChatGPT-referred traffic just hit a 12-month low.
What to do: pull your category's AEO Sensor view and brief one team on the citation gap.Alibaba wired Qwen into Taobao and Tmall on May 11 with a conversational shopping agent, virtual try-on, and 30-day price tracking. "Chat-to-buy" is now China's default e-commerce thesis.
What to do: if you sell on Tmall, draft an agent-readable feed plan this quarter.Google's biggest marketers' keynote hits May 20 with AI-powered campaigns, agentic commerce, and YouTube performance. Meridian Studio + GeoX sit underneath as the measurement story.
What to do: write a one-page "what we'll pilot in Q3" memo before Wednesday's keynote.Walmart Connect — already the fastest-growing retail-media network on earth, +41% YoY in Q4 — shipped two AI moves this month. Sparky, the customer-facing assistant 81% of Walmart shoppers say they've used, now hosts sponsored ads inside the agentic flow. Marty, the advertiser-side AI, rolls to every Sponsored Search brand this half. One customer agent recommends your product. One advertiser agent builds the campaign promoting it.
Three concrete moves a marketer could make. If you do nothing else from this issue, consider these.
Test it with "best [your category] under $X" and "compare us vs. [competitor]." Fix the three gaps that surface. The agent reads your bullets like training data — clean copy beats ad spend.
Brief one team on the citation-share gap between you and your top three competitors. AI visibility just became measurable at industry level — for free. Your competitors will brief from it this week.
Before Google's May 20 keynote, pick one agentic-ad or Meridian surface you'll test in Q3 and name an owner. The keynote sets the planning floor for H2 — get ahead of the dust.
This is not an assignment. It's for your use as you see fit — read it, skip it, forward it, file it. The Playbook is a set of options, not a checklist.
Michigan preliminary sentiment 48.2, CPI 3.3%, 30-yr mortgage 6.37% — the customer your AI is selling to is the most squeezed since 2009.