<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Kenakamu — Writing</title>
    <link>https://kenakamu.org/writing.html</link>
    <atom:link href="https://kenakamu.org/feed.xml" rel="self" type="application/rss+xml"/>
    <description>Write-ups of problems hit in real client work: .NET, Azure, integrations and LLM applications. No industry news.</description>
    <language>en</language>
    <item>
      <title>Why your RAG answers confidently and wrongly</title>
      <link>https://kenakamu.org/rag-answers-confidently-and-wrongly.html</link>
      <guid isPermaLink="true">https://kenakamu.org/rag-answers-confidently-and-wrongly.html</guid>
      <pubDate>Fri, 12 Jun 2026 09:00:00 +0000</pubDate>
      <description>Nearly every failing RAG system I have reviewed was not broken in the model. The model did exactly what it does well: fluently retell what it was handed. Where retrieval breaks, why retrieval beats prompting, and how to put evals in CI.</description>
    </item>
  </channel>
</rss>
