<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
<title>KaiZai</title>
<link>https://www.kaizai.co/blog/</link>
<atom:link href="https://www.kaizai.co/blog/feed.xml" rel="self" type="application/rss+xml"/>
<description>Articles from KaiZai.</description>
<language>en-gb</language>
<item><title>Memory is not a context strategy</title><link>https://www.kaizai.co/blog/memory-is-not-a-context-strategy/</link><guid isPermaLink="true">https://www.kaizai.co/blog/memory-is-not-a-context-strategy/</guid><pubDate>Fri, 06 Nov 2026 00:00:29 +0000</pubDate><description>An 8B-parameter model scores 42.8% on GAIA, a hard agent benchmark, while reading 6–7 times fewer input tokens than a standard ReAct agent. It does it by throwing its own context away.</description></item>
</channel>
</rss>
