<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Myco Brain Blog</title>
    <link>https://mycobrain.dev/blog</link>
    <description>Technical posts, case studies, and research from the Myco Brain team.</description>
    <language>en-us</language>
    <atom:link href="https://mycobrain.dev/blog/rss.xml" rel="self" type="application/rss+xml"/>
    <item>
      <title>Building an agent that remembers across 10,000 conversations without hallucination</title>
      <link>https://mycobrain.dev/blog/agent-memory-10000-conversations</link>
      <guid>https://mycobrain.dev/blog/agent-memory-10000-conversations</guid>
      <pubDate>Wed, 03 Jun 2026 00:00:00 GMT</pubDate>
      <description>How we built a memory system that scales to tens of thousands of agent interactions. Deterministic writes, confidence gating, and graph traversal keep retrieval precise as the corpus grows.</description>
    </item>
    <item>
      <title>Why we store agent memories in a knowledge graph, not a vector database</title>
      <link>https://mycobrain.dev/blog/knowledge-graph-not-vector-db</link>
      <guid>https://mycobrain.dev/blog/knowledge-graph-not-vector-db</guid>
      <pubDate>Wed, 27 May 2026 00:00:00 GMT</pubDate>
      <description>Vector databases are the default for agent memory. We ripped ours out. Here&apos;s why a relational knowledge graph on Postgres outperforms pure vector search for agent recall.</description>
    </item>
    <item>
      <title>The noise problem in AI agent memory — and how we solved it</title>
      <link>https://mycobrain.dev/blog/noise-problem-ai-agent-memory</link>
      <guid>https://mycobrain.dev/blog/noise-problem-ai-agent-memory</guid>
      <pubDate>Wed, 20 May 2026 00:00:00 GMT</pubDate>
      <description>We audited 10,134 agent memory entries and found 97% were noise. Why this happens, why vector databases make it worse, and the deterministic system that fixes it.</description>
    </item>
    <item>
      <title>97% of agent memories were noise: methodology and reproducibility</title>
      <link>https://mycobrain.dev/blog/97-percent-noise</link>
      <guid>https://mycobrain.dev/blog/97-percent-noise</guid>
      <pubDate>Wed, 20 May 2026 00:00:00 GMT</pubDate>
      <description>We audited 10,134 memory entries and labeled 9,791 as noise — duplicates, hallucinations, or malformed extractions. Here&apos;s how we measured it and how you can reproduce it.</description>
    </item>
  </channel>
</rss>