<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>Pulsar blog</title>
    <link>https://starlight-search.com/blog</link>
    <atom:link href="https://starlight-search.com/blog/rss.xml" rel="self" type="application/rss+xml" />
    <description>Notes from the Pulsar team on running AI agents at work.</description>
    <language>en</language>
    <item>
      <title>From traces to know-how: procedural memory for AI agents</title>
      <link>https://starlight-search.com/blog/agentrtx-procedural-memory-ai-agents</link>
      <guid isPermaLink="true">https://starlight-search.com/blog/agentrtx-procedural-memory-ai-agents</guid>
      <pubDate>Tue, 17 Nov 2026 09:00:00 GMT</pubDate>
      <dc:creator>Sonam Pankaj</dc:creator>
      <category>agentRTX</category>
      <description>Tracing tools record what your agents did. agentRTX turns those runs and their corrections into procedural memory, skills and better models. τ-bench: 56% → 80%.</description>
    </item>
    <item>
      <title>Muse, Grok Bot and Pulsar: what changes when an agent works for a company</title>
      <link>https://starlight-search.com/blog/muse-vs-grok-bot-vs-pulsar</link>
      <guid isPermaLink="true">https://starlight-search.com/blog/muse-vs-grok-bot-vs-pulsar</guid>
      <pubDate>Tue, 17 Nov 2026 09:00:00 GMT</pubDate>
      <dc:creator>Sonam Pankaj</dc:creator>
      <category>Governance</category>
      <description>How Muse, Grok Bot and Pulsar differ on tool permissions, data retention and private cloud, and why a company agent should remember what your team knows.</description>
    </item>
  </channel>
</rss>
