Tag
Windows 11的最新Insider预览版加入了Xbox游戏模式下的快捷设置（无需退出游戏即可控制蓝牙、WiFi和电源），并大幅改进了系统搜索功能，包括结果筛选、直接执行操作以及可训练的搜索反馈机制。
Trakkr Research tested 1,285 statistics cited by ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews, finding only 40% traced back to a study saying the same thing, while many 'zombie stats' circulate with no research behind them.
The author of SocialCrawl introduces 'Discover', a cross-platform social listening API endpoint for AI agent builders that searches multiple platforms, scores posts against natural-language questions, dedupes, and returns a unified result list with spend controls. Feedback and free API credits are being offered for testers.
Context.dev 宣布完成 900 万美元融资（由 General Catalyst 领投），并推出面向 AI 智能体的统一 Web 数据 API，集搜索、抓取、爬取与页面监控于一体，帮助开发者用一个密钥和一份账单替代多家供应商。
SearchTok is a new tool that makes TikTok and Instagram content searchable for AI agents, allowing agents to query social media data.
Quello 在 Product Hunt 上线，定位为统一工作上下文的 AI 知识工作区，通过 MCP 将文件、工具与文档的完整上下文带入 Claude、ChatGPT、Gemini 等模型，并提供带引用的答案与语音助手。
Science MadeReadable introduces an algorithmic expert-discovery feature that helps users and AI agents find leading researchers in a field, demonstrated by identifying Richard Sutton as a top Reinforcement Learning expert.
L2G-SR 提出用学习到的结构验证器（基于改进的树编辑距离）引导 LLM 进行符号回归搜索，弥补纯数值误差无法识别正确公式结构的不足，并在 LLM-SRBench 上取得最先进的数值拟合与结构准确率。
UC Berkeley researchers show that scaling LLM research agents stalls when agents anchor on their own history, and propose periodically discarding session state and restarting from the best candidate — a 'forgetting' approach that yields up to 15% gains over parallel agents and 39% over a single agent at the same budget.
Google is updating its AI search results to make inline links more prominent in AI Overviews and AI Mode, aiming to address concerns about reduced traffic to publishers, while also expanding its News AI licensing pilot to over 400 publishers.
Sergio Pineiro 惊叹某开源搜索工具能在 30 毫秒内检索 3200 万条记录，性能极其惊人，并且设置简单、支持模糊搜索、向量搜索和语义搜索。
LSearch 是一个 Windows 上的 Spotlight 风格搜索工具，使用 EmbeddingGemma 2 (270M) 嵌入模型，并组合了两个微调的 Qwen 3.5 4B 模型（分别针对编码和 agentic 任务）构成 MoE，用于代码修复。
一位网友在 GitHub 上发现了一款基于 Google EmbeddingGemma2 小模型的 Mac 搜索应用，称其准确度远超系统自带 Spotlight，支持文件名、文件内容、图片和音频搜索。
一篇实验性博客文章提出「Bag of Decisions」重排方法：让 LLM 为每个查询生成一组是/否判准问题，再用 Jev 计算各问题为真的概率并求和作为相关性分数，在 Wayfair WANDS 和 Amazon ESCI 数据集上均优于 BM25 和单一问题重排。
Microsoft 正在测试全新设计的 Windows 11 搜索功能，基于 WinUI 3 构建，速度更快、支持内联结果预览，并可作为启动器执行开关蓝牙、发短信等操作，预计未来几个月内向正式版推出。
Researchers introduce EPOCH, an evidence-governed architecture for AI research agents that combines task contracts, typed memory, active falsification, and independent replay, achieving state-of-the-art performance on AlgoTune (0.65 vs 0.53 mean normalized score) and the Math14 suite.
Google DeepMind unveiled a 740M-parameter multimodal model that is competitive across benchmarks, outperforming some specialist models more than twice its size, enabling developers to add multimodal search features like finding video moments via voice memo and pair it with Gemma 4 for private on-device use.
A blogger used Codex to search X for high-quality AI 3D creators and found themselves ranked third, reinforcing their focus on AI 3D research and plans to publish AI 3D tutorials after National Day.
This paper studies how LLM agents exhibit source preference in end-to-end search, favoring items from certain sites even when they meet fewer requirements, and demonstrates that supplying missing information or counter-prompting can reduce this bias across 12 agent models.
Turbopuffer announced a major storage architecture change ("turbopuffer v3") that demotes the ANN vector index from primary to secondary index, evolving the serverless database from a specialized vector search engine into a generalized search database supporting many more query plans at greater scale.