<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>MATRIX · ThinkingBox</title>
  <link>https://derekwang2002.github.io/ai-intelligence/en/projects/thinkingbox/</link>
  <description>Material changes in AI models, agents, engineering and infrastructure, with primary sources and recommendations.</description>
  <language>en</language>
  <lastBuildDate>Mon, 02 Nov 2026 16:56:00 GMT</lastBuildDate>
  <atom:link href="https://derekwang2002.github.io/ai-intelligence/en/feeds/projects/thinkingbox.xml" rel="self" type="application/rss+xml"/>
  <item>
    <title>Microsoft opens ThinkingBox to check agents against final database state</title>
    <link>https://derekwang2002.github.io/ai-intelligence/en/events/ev-20261003-02/</link>
    <guid isPermaLink="false">ev-20261003-02-published</guid>
    <pubDate>Sun, 04 Oct 2026 16:02:19 GMT</pubDate>
    <category>New</category>
    <category>TRIAL</category>
    <description>&lt;p&gt;Microsoft and Hugging Face made ThinkingBox runnable through OpenEnv. Its 507 stateful business workflows execute 20 trials per model and grade terminal database state and side effects rather than trusting an agent transcript. Code and data are public, and the common-set analysis covers 121,680 valid trials.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Why read: &lt;/strong&gt;ThinkingBox checks terminal database state, exposing silent failures transcript graders miss.&lt;/p&gt;&lt;p&gt;Recommendation: TRIAL&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://huggingface.co/blog/microsoft/thinkingbox&quot;&gt;Microsoft and Hugging Face: ThinkingBox&lt;/a&gt; · &lt;a href=&quot;https://github.com/microsoft/thinkingbox&quot;&gt;Microsoft ThinkingBox&lt;/a&gt; · &lt;a href=&quot;https://github.com/microsoft/thinkingbox-data/releases/tag/thinkingbox-bench-v1.0&quot;&gt;ThinkingBox Bench v1.0 data&lt;/a&gt;&lt;/p&gt;</description>
  </item>
</channel>
</rss>
