<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Axiom AI blog</title>
    <link>https://frosoore.github.io/AxiomAI/blog/index.html</link>
    <atom:link href="https://frosoore.github.io/AxiomAI/feed.xml" rel="self" type="application/rss+xml" />
    <description>News, deep dives and the life of the Axiom AI project.</description>
    <language>en</language>
    <lastBuildDate>Tue, 23 Jun 2026 00:00:00 +0000</lastBuildDate>
    <item>
      <title>We asked an AI to tear Axiom apart</title>
      <link>https://frosoore.github.io/AxiomAI/blog/we-asked-an-ai-to-tear-axiom-apart.html</link>
      <guid isPermaLink="true">https://frosoore.github.io/AxiomAI/blog/we-asked-an-ai-to-tear-axiom-apart.html</guid>
      <pubDate>Tue, 23 Jun 2026 00:00:00 +0000</pubDate>
      <description>A stranger gave me a tip: feed your project to a capable AI and tell it to attack your assumptions, not flatter them. So we did. / The uncomfortable finding: our 'deterministic firewall' guards the game-state numbers, not the story text, so the prose can still fib for a turn. / Nothing here is fatal. Two fixes plus our existing test-harness and actor-model plans just moved to the top of the roadmap.</description>
    </item>
    <item>
      <title>Under the hood: how saves and universes work</title>
      <link>https://frosoore.github.io/AxiomAI/blog/how-saves-and-universes-work.html</link>
      <guid isPermaLink="true">https://frosoore.github.io/AxiomAI/blog/how-saves-and-universes-work.html</guid>
      <pubDate>Sun, 21 Jun 2026 00:00:00 +0000</pubDate>
      <description>We ran a full quality pass on the save and universe system, so here's a tour of how the thing actually works. / Your universe is plain text you can read and version, the database is just a throwaway cache, and each save is one self-contained, shareable file. / The QC caught a sneaky rewind bug, we fixed it, and we added a test so it can never sneak back.</description>
    </item>
    <item>
      <title>What we took from Hindsight</title>
      <link>https://frosoore.github.io/AxiomAI/blog/what-we-took-from-hindsight.html</link>
      <guid isPermaLink="true">https://frosoore.github.io/AxiomAI/blog/what-we-took-from-hindsight.html</guid>
      <pubDate>Sat, 20 Jun 2026 00:00:00 +0000</pubDate>
      <description>Axiom's memory got a serious rework, inspired by an open-source project called Hindsight. / NPCs can now keep facts, form opinions that shift over time, and hold a living profile of who they are. / It's all optional, and the offline mode that never touches an AI still works exactly like before.</description>
    </item>
    <item>
      <title>Axiom has a blog now</title>
      <link>https://frosoore.github.io/AxiomAI/blog/the-blog-is-live.html</link>
      <guid isPermaLink="true">https://frosoore.github.io/AxiomAI/blog/the-blog-is-live.html</guid>
      <pubDate>Sat, 20 Jun 2026 00:00:00 +0000</pubDate>
      <description>We finally have a real place to talk about the project, not just a dry changelog. / Expect deep dives, the thinking behind the big changes, and the messes we walk into. / There's an RSS feed too, so you can follow along without ever opening the site.</description>
    </item>
  </channel>
</rss>
