<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Ai-Infrastructure on BRYSGO</title><link>https://www.brysgo.com/tags/ai-infrastructure/</link><description>Recent content in Ai-Infrastructure on BRYSGO</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 14 Jul 2026 03:14:36 +0000</lastBuildDate><atom:link href="https://www.brysgo.com/tags/ai-infrastructure/index.xml" rel="self" type="application/rss+xml"/><item><title>Your Agent Loop Has a Compounding Cost Problem</title><link>https://www.brysgo.com/post/2026-07-14-your-agent-loop-has-a-compounding-cost-problem/</link><pubDate>Tue, 14 Jul 2026 03:14:36 +0000</pubDate><guid>https://www.brysgo.com/post/2026-07-14-your-agent-loop-has-a-compounding-cost-problem/</guid><description>&lt;p&gt;A 30-step AI workflow doesn&amp;rsquo;t cost 30 times a single API call. It costs closer to 465 times, because the model re-reads every previous message on every single turn. That&amp;rsquo;s not a quirk you can optimize around at the margins. It&amp;rsquo;s baked into how stateless LLM APIs work, and if you&amp;rsquo;re building anything non-trivial with agents, it&amp;rsquo;s quietly eating your budget in a way that compound interest would recognize immediately.&lt;/p&gt;</description></item></channel></rss>