<?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>Ai on NavShelf Blog</title>
    <link>https://navshelf.com/en/tags/ai/</link>
    <description>Recent content in Ai on NavShelf Blog</description>
    <generator>Hugo</generator>
    <language>en</language>
    <lastBuildDate>Sun, 13 Sep 2026 11:00:00 +0800</lastBuildDate>
    <atom:link href="https://navshelf.com/en/tags/ai/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Why Your AI Bill Is Harder to Predict Than Your Hosting Bill</title>
      <link>https://navshelf.com/en/posts/ai-cost-predictable/</link>
      <pubDate>Sun, 13 Sep 2026 11:00:00 +0800</pubDate>
      <guid>https://navshelf.com/en/posts/ai-cost-predictable/</guid>
      <description>Hosting bills scale with traffic; token bills scale with context and behaviour. Where AI spend actually comes from, which levers move it, and how to estimate a workload before it ships — with a free calculator that never fetches prices for you.</description>
    </item>
    <item>
      <title>AI Coding Agents Need Guardrails, Not More Trust</title>
      <link>https://navshelf.com/en/posts/ai-agent-guardrails/</link>
      <pubDate>Sun, 13 Sep 2026 10:00:00 +0800</pubDate>
      <guid>https://navshelf.com/en/posts/ai-agent-guardrails/</guid>
      <description>An agent that edits forty files while you are reviewing the first one is not a tooling problem, it is a missing-policy problem. How to encode the rules your team already agrees on as a rules file, permission lists, protected paths and a review checklist.</description>
    </item>
    <item>
      <title>How to Check Whether an AI Assistant Made Up the Numbers in Its Answer</title>
      <link>https://navshelf.com/en/posts/verify-ai-numbers/</link>
      <pubDate>Sun, 13 Sep 2026 09:00:00 +0800</pubDate>
      <guid>https://navshelf.com/en/posts/verify-ai-numbers/</guid>
      <description>AI answers contain numbers that look right and are not: invented durations, invented percentages, totals that never appeared in the export. Here is a five-minute workflow for checking AI output against the raw data, and a free checker that flags untraceable numbers.</description>
    </item>
  </channel>
</rss>
