<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" version="2.0">
  <channel>
    <title>Blog (Momentuum)</title>
    <link>https://www.momentuumblue.si/blogs</link>
    <description>Blogs</description>
    <language>en</language>
    <pubDate>Thu, 19 Nov 2026 15:24:02 GMT</pubDate>
    <dc:date>2026-11-19T15:24:02Z</dc:date>
    <dc:language>en</dc:language>
    <item>
      <title>Eval Harness for AI: Reproducible, Costed, Traceable Results</title>
      <link>https://www.momentuumblue.si/blogs/evaluation-harness-for-ai-reproducible-costed-traceable-results</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://www.momentuumblue.si/blogs/evaluation-harness-for-ai-reproducible-costed-traceable-results" title="" class="hs-featured-image-link"&gt; &lt;img src="https://www.momentuumblue.si/hubfs/AI-Generated%20Media/Images/Modern%20Tech%20Workspace%20With%20City%20View.png" alt="Eval Harness for AI: Reproducible, Costed, Traceable Results" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;Every enterprise faces recurring questions that require correct and consistent answers. This was simpler when all enterprise code performed deterministic tasks. With the rise of AI and agentic workflows, tasks and responses have become more probabilistic, and the same metrics no longer work for agents. This is because your agents follow orders well. In fact, they are so good that they’ll invent facts (“hallucinate”) to produce a viable answer rather than say they don’t know. These hallucinations and errors have causes, but you can’t see them in the output because a fabricated answer can appear just as confident as a correct one.&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://www.momentuumblue.si/blogs/evaluation-harness-for-ai-reproducible-costed-traceable-results" title="" class="hs-featured-image-link"&gt; &lt;img src="https://www.momentuumblue.si/hubfs/AI-Generated%20Media/Images/Modern%20Tech%20Workspace%20With%20City%20View.png" alt="Eval Harness for AI: Reproducible, Costed, Traceable Results" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;Every enterprise faces recurring questions that require correct and consistent answers. This was simpler when all enterprise code performed deterministic tasks. With the rise of AI and agentic workflows, tasks and responses have become more probabilistic, and the same metrics no longer work for agents. This is because your agents follow orders well. In fact, they are so good that they’ll invent facts (“hallucinate”) to produce a viable answer rather than say they don’t know. These hallucinations and errors have causes, but you can’t see them in the output because a fabricated answer can appear just as confident as a correct one.&lt;/p&gt;  
&lt;img src="https://track-eu1.hubspot.com/__ptq.gif?a=25186482&amp;amp;k=14&amp;amp;r=https%3A%2F%2Fwww.momentuumblue.si%2Fblogs%2Fevaluation-harness-for-ai-reproducible-costed-traceable-results&amp;amp;bu=https%253A%252F%252Fwww.momentuumblue.si%252Fblogs&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Artificial Intelligence</category>
      <category>Generative AI</category>
      <category>AI Engineering</category>
      <category>Machine Learning Operations</category>
      <category>AI Evaluation &amp; Testing</category>
      <pubDate>Thu, 19 Nov 2026 15:23:32 GMT</pubDate>
      <guid>https://www.momentuumblue.si/blogs/evaluation-harness-for-ai-reproducible-costed-traceable-results</guid>
      <dc:date>2026-11-19T15:23:32Z</dc:date>
      <dc:creator>Kanit Mann</dc:creator>
    </item>
  </channel>
</rss>
