<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Engineering on Giulio Calacoci</title><link>https://gcalacoci.com/categories/engineering/</link><description>Recent content in Engineering on Giulio Calacoci</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Mon, 16 Nov 2026 02:30:00 +0100</lastBuildDate><atom:link href="https://gcalacoci.com/categories/engineering/index.xml" rel="self" type="application/rss+xml"/><item><title>Deterministic checks around non-deterministic agents</title><link>https://gcalacoci.com/posts/deterministic-checks-around-non-deterministic-agents/</link><pubDate>Mon, 16 Nov 2026 02:30:00 +0100</pubDate><guid>https://gcalacoci.com/posts/deterministic-checks-around-non-deterministic-agents/</guid><description>&lt;p&gt;A model can write good code and then write bad code from the same prompt. I run agents on real work every day, so I can&amp;rsquo;t rely on the model behaving. My approach is to leave the model free to vary and make everything around it fixed: the task, the checks, the routing, the logs and who approves what. This note describes how that works in my own agent systems.&lt;/p&gt;</description></item></channel></rss>