<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Methodology on Local First AI</title><link>https://localfirstai.eu/tags/methodology/</link><description>Recent content in Methodology on Local First AI</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Wed, 24 Jun 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://localfirstai.eu/tags/methodology/index.xml" rel="self" type="application/rss+xml"/><item><title>We Reviewed Our Own Legal Brief with an Adversarial AI Panel. Zero of Seven Claims Survived Unchanged.</title><link>https://localfirstai.eu/posts/2026-06-24-adversarial-legal-panel/</link><pubDate>Wed, 24 Jun 2026 00:00:00 +0000</pubDate><guid>https://localfirstai.eu/posts/2026-06-24-adversarial-legal-panel/</guid><description>A single AI model drafting a legal brief produces confident prose. Three AI models with conflicting mandates produce a brief that survives scrutiny. Here&amp;#39;s the methodology and what the panel found.</description></item></channel></rss>