<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Token-Efficiency on Advancing Engineering</title><link>https://www.advancingengineering.dev/tags/token-efficiency/</link><description>Recent content in Token-Efficiency on Advancing Engineering</description><generator>Hugo</generator><language>en</language><lastBuildDate>Mon, 05 Oct 2026 13:21:22 +0100</lastBuildDate><atom:link href="https://www.advancingengineering.dev/tags/token-efficiency/index.xml" rel="self" type="application/rss+xml"/><item><title>Stop Paying for Noise: RTK Brings Token Efficiency Naturally</title><link>https://www.advancingengineering.dev/posts/2026-09-rtk-token-killer/</link><pubDate>Fri, 25 Sep 2026 11:50:20 +0100</pubDate><guid>https://www.advancingengineering.dev/posts/2026-09-rtk-token-killer/</guid><description>&lt;p&gt;Every tool call an agent makes still behaves like a person sitting at a terminal. It runs &lt;code&gt;git log&lt;/code&gt; and gets back commit hashes, author names, dates, and full commit bodies. Run a test suite and it hands back a progress bar and a line for every test that passed. All of that was built for a human&amp;rsquo;s eyes, skimming past most of it to the two lines that matter.&lt;/p&gt;
&lt;p&gt;An agent doesn&amp;rsquo;t skim, it reads every word, and you pay for the privilege.&lt;/p&gt;
&lt;p&gt;I started running tool calls through RTK this week, and after a couple of thousand uses it&amp;rsquo;s tracking says 5.1 million tokens have been saved. For week one that&amp;rsquo;s a saving of &lt;strong&gt;$10-$20&lt;/strong&gt;, depending on the models used. Stretch that across the rest of the year and it could be over $1,000.&lt;/p&gt;
&lt;p&gt;Of course, our total AI spend is bigger than tool-call tokens alone, but saving &lt;strong&gt;$1,000&lt;/strong&gt; a year for one install is good.&lt;/p&gt;</description></item></channel></rss>