Agent Beat for October 18, 2026

Welcome to Agent Beat, the daily briefing on how AI finds brands and how agents choose what to buy. Today is Sunday, October 18, 2026. We begin with search visibility and the signals Google is now putting in front of site operators.

Google is emailing site operators to tell them how many users have chosen the site as a preferred source, SEO Südwest reports. The email gives site operators a count of users who made that choice. Preferred Sources decide which sites Google shows first in Top Stories, AI Overviews and AI Mode. That gives site operators a direct count for a setting that affects placement in three Google surfaces.

Meta has also changed language that governs how its services interact with the web. Meta removed wording from its terms that had raised concern about web scraping and AI training, TechCrunch reports. A Meta spokesperson told the outlet that the change did not alter how Meta crawls the web. Earlier coverage said the terms widened the rights ad buyers had granted. The change concerns the terms themselves, while the spokesperson's statement concerns crawling.

The way researchers measure brand visibility is under review too. A paper tested whether Google search results and OpenAI's API can stand in for ChatGPT's own citations when researchers measure brand visibility. The test compared those stand-ins with ChatGPT's own citations. The paper found only a partial match. The cited sites matched more often than the exact pages did, according to Search Engine Journal's write-up.

The tools researchers use to measure citations are changing alongside search products. DeepSeek opened its own search engine API on October 17 at 75 cents per 1,000 successful search requests, according to the article. The article says the API has a Search endpoint for web queries and a Fetch endpoint that formats pages for AI reading. It also says DeepSeek will filter SEO cheating, GEO manipulation and low-quality AI-generated content. The article names Tencent, ByteDance, Alibaba, Zhipu and Kimi as having built search products. It frames the shift as pressure on Baidu.

The pressure is not limited to search engines. Advertising is changing around the same systems. A MarkeZine column describes ads inside ChatGPT, ads run by brand agents, ad consoles operated by agents, and a reshuffle in how ad transactions are connected as changes from the third quarter of 2026. The column adds practical advice for marketers facing more automated campaign management.

German advertisers have their own adjustments to make. The article argues ChatGPT Ads carries over performance marketing basics but does not work like Google Ads in a chat window. Conversation context and user intent matter more than single keywords, it says. The article's comparison is between ChatGPT Ads and Google Ads. It notes ChatGPT Ads have been available in Germany since August 24, 2026, with self-service access through the Ads Manager. It says U.S. results are not yet reliable benchmarks for Germany.

Beyond the ad auction, the purchase decision itself is getting attention. A MarkeZine report from MarkeZine Day 2026 Autumn covers a session with Daiko and Laboro.AI on making a brand's value legible to AI. The session covered implementing an ontology, a structured set of terms, so a model describes the company correctly. It also covered designing communications around five patterns of AI use at purchase time. The session started from the observation that asking ChatGPT about a company returns a mostly correct answer that does not by itself make a buyer choose it.

ChatGPT is also moving closer to the purchase itself. ChatGPT added a virtual try-on for clothing and a favorites feature for saving products, t3n reports. The try-on builds a preview from an uploaded selfie. The report frames the addition as ChatGPT becoming a virtual fashion adviser.

With more of that shopping happening through AI, budget planning is shifting too. iPullRank argues that end-of-year budget spent on AI search buys measurement infrastructure, technical fixes and content built to be retrieved and cited. That is different from paid media that stops when the spend stops. The argument concerns Q4 budget. The post names fear and vendor scores that disagree with each other as what keeps budgets out of the channel.

Local brands are working through the same questions in their own markets. A Journal du Net piece says four practitioners give their recipes for local brand visibility in AI chat engines. The piece focuses on local brand visibility in conversational search. It says AI sends more qualified local traffic while classic SEO work still produces strong results. It also says the channel is volatile and needs a well-established local presence.

That is the briefing for Sunday, October 18, 2026. Thank you for listening. For the full written brief, go to agent-beat.com.