<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>ARMOUR</title><link>https://armour.ai/</link><description>Recent content on ARMOUR</description><generator>Hugo</generator><language>en-gb</language><lastBuildDate>Sun, 24 May 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://armour.ai/index.xml" rel="self" type="application/rss+xml"/><item><title>The Lossy Chain</title><link>https://armour.ai/blog/the-lossy-chain/</link><pubDate>Sun, 24 May 2026 00:00:00 +0000</pubDate><guid>https://armour.ai/blog/the-lossy-chain/</guid><description>&lt;p&gt;I told my coding agent to remember something important: the game engine uses C++, and GDScript is only for front-end code. It dutifully wrote this down in a memory file and confirmed it would remember. The next day, I asked it to implement an engine feature. It happily coded the whole thing in GDScript.&lt;/p&gt;
&lt;p&gt;When I asked why, it told me it didn't know I wanted C++. &amp;quot;Look in your memory,&amp;quot; I said. &amp;quot;Oh, there it is! I'm so sorry. I'll start over in C++.&amp;quot;&lt;/p&gt;</description></item><item><title>About</title><link>https://armour.ai/about/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://armour.ai/about/</guid><description>&lt;p&gt;I'm AJ Rogers. I build agent memory systems and write about why they don't work yet.&lt;/p&gt;
&lt;p&gt;This blog is a public lab notebook. I'm working through the problems of long-term memory for AI agents--how to store it, how to retrieve it, how to make sure the model actually uses it when it matters. The field is young and most of the interesting problems are unsolved.&lt;/p&gt;
&lt;p&gt;I've spent the last year building memory infrastructure for coding agents, watching it fail in new and interesting ways, and trying to figure out what a real solution looks like. These posts are the notes from that process.&lt;/p&gt;</description></item></channel></rss>