<?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>Research Notes on YanlunAI</title><link>https://yanlunai.com/en/categories/research-notes/</link><description>Recent content in Research Notes on YanlunAI</description><generator>Hugo -- 0.152.2</generator><language>en</language><lastBuildDate>Mon, 15 Dec 2025 20:10:00 +0800</lastBuildDate><atom:link href="https://yanlunai.com/en/categories/research-notes/index.xml" rel="self" type="application/rss+xml"/><item><title>From chat to agents: planning is the dividing line</title><link>https://yanlunai.com/en/posts/papers/2025-12-15-agent-planning-view/</link><pubDate>Mon, 15 Dec 2025 20:10:00 +0800</pubDate><guid>https://yanlunai.com/en/posts/papers/2025-12-15-agent-planning-view/</guid><description>An agent is more than a multi-turn chat interface: it maintains state, decomposes goals, uses tools, and recovers from failure.</description></item><item><title>An engineering view of trustworthy RAG</title><link>https://yanlunai.com/en/posts/papers/2025-12-15-clean-rag-engineering/</link><pubDate>Mon, 15 Dec 2025 17:40:00 +0800</pubDate><guid>https://yanlunai.com/en/posts/papers/2025-12-15-clean-rag-engineering/</guid><description>Trustworthy RAG is not a prompt. It is a system for traceable evidence, grounded answers, and abstention under uncertainty.</description></item></channel></rss>