AI Search news and analysis on Web Pulse: 728 curated articles tracking the latest AI Search developments, tools, and research, updated continuously from vetted sources.
New research titled "Data-Void Exploits in Web-Searching LLMs" (arXiv 2610.30365) shows that attackers can seed a single convincing webpage in a data void—a query where credible coverage is scarce—so a web-searching LLM …
Manticore Software built the open-source Manticore Search Playground, a seven-tab demo that runs full-text, vector, hybrid, image, geo, comparison, and AI question-answering search against a single catalog of 82,524 prod…
In a 40-query comparison of four search APIs built for AI agents, Exa returned the best search results for 31 of the 40 queries, while Firecrawl returned usable page content from 37 of 40 pages, including all 10 PDFs, an…
A developer published a step-by-step method for verifying whether AI crawlers such as GPTBot, ClaudeBot and PerplexityBot can actually fetch a site, starting with the robots.txt file and confirming real behavior through …
A developer published a seven-point release checklist for building candidate-search retrieval systems that combine a tenant-scoped vector index with live web search, arguing that retrieval policy—not the LLM—determines w…
Envoyra, a tool that checks whether AI assistants recommend a brand, ran its own product through ChatGPT, Claude, Gemini and Perplexity on 21 October 2026, asking 18 questions once each for 72 total answers. The assistan…
Microsoft launched Topics Portfolio, a beta feature in Bing Webmaster Tools that clusters grounding queries into broader themes and flags where other sites are cited in AI search responses but the user's site is not. The…
A developer published a 10-minute curl-based test for checking whether AI crawlers can see page content that only appears after JavaScript runs. The method fetches a URL with curl, greps for key facts like headlines and …
Atalaya, a company run by AI agents with a human owner, analyzed the robots.txt files of 751 popular sites across Mexico, the US and Canada against 20 AI crawlers in October 2026, finding that 20.2% of Mexican sites bloc…
Pinecone has added integrated embedding to its Documents API, letting one text field serve both semantic and full-text search with embedding and search billed together on a single Pinecone invoice. Three embedding models…
Pinecone added integrated embedding to its Documents API, so developers can send raw text and have Pinecone run the embedding pipeline on every write and every search, the company said in a blog post. The move follows Pi…
SerpApi reported that returning search results as Markdown instead of JSON can cut AI SEO analysis token costs by up to 90%, because Markdown tables state field names once in a header rather than repeating keys such as p…
A developer published a robots.txt walkthrough for WordPress sites that separates search crawlers from AI training crawlers, allowing Googlebot, Bingbot and similar bots while disallowing GPTBot, CCBot, ClaudeBot, Google…
Fact discovery closed on October 15, 2026, in Dow Jones & Company v. Perplexity AI (case 1:24-cv-07984) in the Southern District of New York, with expert discovery due by January 19, 2027 and both sides estimating a 10 t…
BlueFacts launched a remote Model Context Protocol server at https://bluefacts.app/mcp that connects AI assistants including Claude Code, Claude Desktop, Cursor, VS Code, Codex and Gemini CLI to daily Bluesky statistics …
A developer published a Python script that approximates Google's query fan-out by using Claude to generate likely sub-queries for a seed question, then splitting a target page into passages, embedding them with a small o…
Ansh Khandelwal, co-founder and strategist at GEO/AEO consultancy Newtation, outlined a measurement framework for tracking whether AI assistants recommend a business, arguing that observation rules should be fixed before…
A new arXiv paper (2610.29250v1) introduces SearchWeaver, a framework that automatically optimizes the interface a language-model agent uses to search a corpus, guided by downstream search performance rather than hand-tu…
A study of 11 retrieval configurations on 396 QASPER questions and 80 questions from eight complete textbooks found that Hybrid + Jev achieved the highest observed evidence coverage on books at 94.1%, versus 88.7% for Je…
A developer built InBrief, a dedupe-first AI news aggregator that pulls from more than 370 sources — RSS, Hacker News, Reddit, X, YouTube transcripts, GitHub Trending and the Hugging Face papers leaderboard — and runs it…