<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Policy Optimization on Programmer.ie: Modern AI programming</title>
    <link>http://programmer.ie/tags/policy-optimization/</link>
    <description>Recent content in Policy Optimization on Programmer.ie: Modern AI programming</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Sun, 09 Aug 2026 10:49:00 +0100</lastBuildDate>
    <atom:link href="http://programmer.ie/tags/policy-optimization/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Advanced Agents From First Principles 15: How Do You Optimize an Agent Policy Without Turning It Into Another Black Box?</title>
      <link>http://programmer.ie/post/advanced-agents-from-first-principles-15/</link>
      <pubDate>Sun, 09 Aug 2026 10:49:00 +0100</pubDate>
      <guid>http://programmer.ie/post/advanced-agents-from-first-principles-15/</guid>
      <description>&lt;h1 id=&#34;how-do-you-optimize-an-agent-policy-without-turning-it-into-another-black-box&#34;&gt;How Do You Optimize an Agent Policy Without Turning It Into Another Black Box?&lt;/h1&gt;&#xA;&lt;p&gt;By now our advanced agent can do a lot.&lt;/p&gt;&#xA;&lt;p&gt;It can:&lt;/p&gt;&#xA;&lt;ul&gt;&#xA;&lt;li&gt;route tasks to different models or specialists,&lt;/li&gt;&#xA;&lt;li&gt;decide whether to search,&lt;/li&gt;&#xA;&lt;li&gt;choose a search budget,&lt;/li&gt;&#xA;&lt;li&gt;decide when to escalate,&lt;/li&gt;&#xA;&lt;li&gt;invoke critics,&lt;/li&gt;&#xA;&lt;li&gt;retry or recover,&lt;/li&gt;&#xA;&lt;li&gt;stop when evidence is strong enough,&lt;/li&gt;&#xA;&lt;li&gt;learn from verified production trajectories,&lt;/li&gt;&#xA;&lt;li&gt;and trace the decisions that produced each outcome.&lt;/li&gt;&#xA;&lt;/ul&gt;&#xA;&lt;p&gt;That creates a new problem.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
