Agent Beat for October 13, 2026

Welcome to Agent Beat. It's Tuesday, October 13, 2026. Your daily briefing on AI search and agent commerce.

Ahrefs data reported by Search Engine Journal shows government health, research, and regulator sites climbing ChatGPT's U.S. citation rankings while explainer pages slipped. The shift favors sites that publish the underlying facts. Marketers should move original data, official documentation, and primary pages to the front of their content plans.

Alibaba's Qwen opened trial applications for its personal agent, Qwen Go, and will admit users in waves. Qwen says the agent handles shopping, travel, and logistics. A reporter's test had it pursue a flight delay claim with customer service, ending in a 70 yuan payout. Brands on Alibaba's marketplaces should expect orders, price comparisons, and refund requests from a Qwen agent acting for the customer.

Google plans to reinvent Discover as an AI concierge with tighter YouTube integration, Search Engine Journal reports. That moves Discover away from a feed of article links toward a surface that decides what to recommend to each user. Marketers should expect placements to depend less on what they publish and more on how Google's AI recommends it.

Alibaba's 1688 says more than 100 million annual buyers and more than 1 million factory suppliers can open the Accio AI commerce workbench. Buyers get a purchasing assistant that searches suppliers in 128 countries and compares prices, plus a manager agent for Taobao, Tmall, Douyin, Pinduoduo, and JD stores. Suppliers get a shop assistant for listings, ads, and customer reception. Keep prices, stock, and listing data accurate, because buying agents will run the search and the inquiry.

The newsletter cites Adobe's finding that 39% of U.S. consumers have used AI for shopping and Salesforce's estimate that AI influenced 20% of purchases during last year's holiday weekend. Only 11% of shoppers let AI buy for them. The assistant's influence sits in the shortlist, built from reviews, videos, and brand searches customers produce after they buy. Marketers should treat those as inputs to the AI shortlist and build a loop that keeps producing them.

Newegg says it integrated Google's Universal Commerce Protocol so its products can be bought inside Google AI Mode, Google Search, and Gemini. It follows Wayfair, Walmart, and Macy's, which have supported UCP since January. Check whether your catalog reaches Google through a UCP feed, because items that do not will not be purchasable inside those AI surfaces.

Seven months after OpenAI began selling ads in ChatGPT, Digiday reports marketers still cannot measure campaigns well enough to commit real budgets. Accuracast's Farhad Divecha said a UK lead campaign produced form fills ChatGPT never logged as conversions, and his agency once counted 20 clicks where the platform reported about 100. Web Guide Partner said conversions can take 24 to 36 hours to appear. Run ChatGPT ads as a test budget until you can reconcile conversions in your own analytics, and expect no demographic or prompt-level reporting.

A Microsoft patent granted on August 25 describes how Bing decides to generate an AI answer. It scores each query for complexity and answers only questions that need several sources combined. The query expands into sub-searches, and a small model scores those results and keeps the best-ranked links for the answer text. Plan for Bing's answer surface to be decided at the level of individual links and sub-queries, so both the page and the terms it can be found for matter.

Chris Andrew, co-founder and CEO of the AI search optimization platform Scrunch, which Sitecore acquired in June 2026, describes five KPIs that replace site traffic and an AXP system for feeding correct information to AI agents. The interview cites a 364% increase in exposure for non-branded queries and results at Akamai. Andrew says Japanese companies mostly monitor their visibility while U.S. teams already spend budget on fixing it. Move from watching AI visibility to a repeated measure-and-fix loop with KPIs that do not depend on referral traffic.

Values.media says it connected to M6 and France Televisions so linear TV sponsorship can be bought by software agents, after an earlier test with TF1, according to managing director Emmanuel Crego. The link lets a buyer's agent handle the transaction through the broadcaster's systems. Settle now which inventory rules, budgets, and data your buying agents may act on without a person in the loop.

An article written by the payments firm Fipto argues that the point of agent payments is not to let AI spend freely but to give it a controlled ability to act. It says stablecoins are the settlement rail that makes agent payments work. Teams preparing agent checkout should treat permission limits and settlement rails as one design decision.

Search Engine Roundtable reports Google has been blocking scrapers of its search results more effectively in recent weeks, and many third-party rank tracking tools still cannot report Search data reliably. Check whether rank or AI visibility dashboard data is complete before acting on it.

Search Engine Journal reports Google job listings describe teams that generate and recommend AI content for Discover, plus a design lead for a proactive concierge experience. Marketers should treat Discover as a surface where Google itself decides what to show, rather than a channel that only ranks publisher articles.

At MarkeZine Day 2026 Autumn, ValueCommerce executive officer Takeshi Amano said AI search summarizes and compares information before a buyer decides. He said companies must design how third parties describe them and not just what they publish themselves. Add third-party description to your plans, because the assistant repeats what others say about the brand as much as what the brand says itself.

A Shopify Japan survey reported by MarkeZine says 52% of Japanese consumers may use AI during the Black Friday and year-end shopping season. The outlet gives no further detail on the survey's method or sample. Plan for year-end campaigns in Japan by checking how your products are described in assistant answers before the season starts.

SEO Südwest reports Google finished its September 2026 spam update, which ran for exactly two weeks. Marketers whose rankings moved in the last two weeks should check whether the change lines up with the update before rewriting pages.

TechCrunch reports Meta released Muse Wonder, described as its first frontier model. It is larger and more capable than the earlier Muse Spark, with higher benchmark scores and better results on complex tasks such as coding. Expect the model in developer tools and agent workflows, but nothing in the release changes how brands are cited in AI answers.

That's today's Agent Beat. Thanks for listening. The full brief is at agent-beat.com.