<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Learning-in-Public on oceanheart.ai</title>
    <link>https://www.oceanheart.ai/tags/learning-in-public/</link>
    <description>Recent content in Learning-in-Public on oceanheart.ai</description>
    <generator>Hugo</generator>
    <language>en-gb</language>
    <lastBuildDate>Thu, 20 Aug 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://www.oceanheart.ai/tags/learning-in-public/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Keeping up with my own agents</title>
      <link>https://www.oceanheart.ai/blog/2026-08-20-keeping-up-with-my-own-agents/</link>
      <pubDate>Thu, 20 Aug 2026 00:00:00 +0000</pubDate>
      <guid>https://www.oceanheart.ai/blog/2026-08-20-keeping-up-with-my-own-agents/</guid>
      <description>&lt;p&gt;Yesterday morning I ran four design rounds on a small research project, bouncing the same questions between two models and carrying each one&amp;rsquo;s answer to the other, back to back. The project is about making agents cheaper to run: pay for inference when a situation is genuinely new, replay recorded decisions when it isn&amp;rsquo;t. The rounds were good. Sharp questions, real tradeoffs, decisions I recognised as correct while I read them. In another window I was applying for jobs. By lunch the design had about thirty settled points and I had agreed to all of them.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
