<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>动态专题 on 焱论AI</title><link>https://yanlunai.com/topics/</link><description>Recent content in 动态专题 on 焱论AI</description><generator>Hugo -- 0.152.2</generator><language>zh-cn</language><lastBuildDate>Mon, 24 Aug 2026 10:00:00 +0800</lastBuildDate><atom:link href="https://yanlunai.com/topics/index.xml" rel="self" type="application/rss+xml"/><item><title>长程 Agent：从一次回答到持续完成任务</title><link>https://yanlunai.com/topics/long-horizon-agents/</link><pubDate>Mon, 24 Aug 2026 10:00:00 +0800</pubDate><guid>https://yanlunai.com/topics/long-horizon-agents/</guid><description>跟踪长程任务中的记忆、状态、工具调用、失败恢复和人工接管。</description></item><item><title>AI Coding 与人机协作</title><link>https://yanlunai.com/topics/ai-coding/</link><pubDate>Sun, 23 Aug 2026 10:00:00 +0800</pubDate><guid>https://yanlunai.com/topics/ai-coding/</guid><description>研究任务委托、验证证据与专业经验如何共同决定编码智能体的真实收益。</description></item><item><title>推理效率与模型路由</title><link>https://yanlunai.com/topics/inference-routing/</link><pubDate>Sat, 22 Aug 2026 10:00:00 +0800</pubDate><guid>https://yanlunai.com/topics/inference-routing/</guid><description>从单一跑分转向任务质量、延迟、吞吐与单位成功成本的综合决策。</description></item><item><title>可信 Agent 与系统安全</title><link>https://yanlunai.com/topics/trustworthy-agents/</link><pubDate>Fri, 21 Aug 2026 10:00:00 +0800</pubDate><guid>https://yanlunai.com/topics/trustworthy-agents/</guid><description>把提示注入、权限边界、工具副作用和高风险动作纳入 Agent 的工程设计。</description></item></channel></rss>