<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Taewan Yoon</title>
    <link>https://twyoon.com/</link>
    <description>Writings by Taewan Yoon</description>
    <language>en</language>
    <lastBuildDate>Fri, 02 Oct 2026 17:42:35 GMT</lastBuildDate>
    <atom:link href="https://twyoon.com/rss.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Who Holds the Loop</title>
      <link>https://twyoon.com/writings/who-holds-the-loop/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/who-holds-the-loop/</guid>
      <pubDate>Tue, 01 Sep 2026 00:00:00 GMT</pubDate>
      <description>I spent a week running GPT models inside Claude Code. Here is the harness that decides everything, OpenAI&apos;s new plugin, and the receipts from every claim I tested.</description>
      <category>Claude Code</category>
      <category>OpenAI Codex</category>
      <category>Agentic AI</category>
      <category>System Design</category>
    </item>
    <item>
      <title>What Documentation Does an Agent Need? 729 CLAUDE.md Trials</title>
      <link>https://twyoon.com/writings/what-documentation-does-an-agent-need/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/what-documentation-does-an-agent-need/</guid>
      <pubDate>Tue, 04 Aug 2026 00:00:00 GMT</pubDate>
      <description>729 blind trials on CLAUDE.md and skill files. The same document scored 87% and 13% depending on nothing but the filesystem underneath it.</description>
      <category>Claude Code</category>
      <category>Anthropic</category>
      <category>Agentic AI</category>
      <category>Context Engineering</category>
      <category>Evaluation</category>
    </item>
    <item>
      <title>What Form Should Knowledge Take?</title>
      <link>https://twyoon.com/writings/what-form-should-knowledge-take/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/what-form-should-knowledge-take/</guid>
      <pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate>
      <description>Where the AI frontier is settling on the form curated knowledge should take as a system, and the narrower role RAG and retrieval now play in serving it.</description>
      <category>Knowledge Management</category>
      <category>Enterprise AI</category>
      <category>Agentic AI</category>
      <category>Data Governance</category>
      <category>Retrieval</category>
    </item>
    <item>
      <title>How I Think About Knowledge in Finance</title>
      <link>https://twyoon.com/writings/knowledge-for-finance/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/knowledge-for-finance/</guid>
      <pubDate>Sun, 21 Jun 2026 00:00:00 GMT</pubDate>
      <description>Finance standardized its data for two decades but rarely kept the reasoning behind decisions. AI makes that reasoning cheaper to find. How firms can manage it.</description>
      <category>Knowledge Management</category>
      <category>Financial Data</category>
      <category>Enterprise AI</category>
      <category>Data Governance</category>
      <category>Agentic AI</category>
    </item>
    <item>
      <title>Building OptiMind: Memory Architecture for Claude Code on Mobile</title>
      <link>https://twyoon.com/writings/files-are-the-memory/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/files-are-the-memory/</guid>
      <pubDate>Sat, 30 May 2026 00:00:00 GMT</pubDate>
      <description>Agentic memory is a harder problem than it looks. Here&apos;s what I learned building a memory-enabled agent around GitHub and Claude Code.</description>
      <category>Agentic Memory</category>
      <category>System Design</category>
      <category>Claude Code</category>
      <category>Agentic AI</category>
      <category>Context Engineering</category>
    </item>
    <item>
      <title>Let the Flywheel Spin on Top of the Translation Layer</title>
      <link>https://twyoon.com/writings/translation-layer/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/translation-layer/</guid>
      <pubDate>Thu, 14 May 2026 00:00:00 GMT</pubDate>
      <description>What 6+ years of building and shipping, and 40+ AI rooms, taught me about earning trust in AI.</description>
      <category>Developer Marketing</category>
      <category>Community</category>
      <category>Product Marketing</category>
    </item>
    <item>
      <title>A Science-Based Skincare Routine</title>
      <link>https://twyoon.com/writings/science-based-skincare-routine/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/science-based-skincare-routine/</guid>
      <pubDate>Thu, 23 Apr 2026 00:00:00 GMT</pubDate>
      <description>People keep asking what skincare products I use. My daily routine, paired with the dermatology and photobiology research that justifies each step.</description>
      <category>Medical Science</category>
      <category>Health</category>
      <category>Routine</category>
    </item>
    <item>
      <title>Tracing the Minds Behind Claude Code</title>
      <link>https://twyoon.com/writings/tracing-the-minds-behind-claude-code/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/tracing-the-minds-behind-claude-code/</guid>
      <pubDate>Mon, 06 Apr 2026 00:00:00 GMT</pubDate>
      <description>I read two Claude Code source trees side by side to trace what changed, and what held, across 13 months of the product&apos;s development.</description>
      <category>Claude Code</category>
      <category>Anthropic</category>
      <category>Agentic AI</category>
      <category>System Design</category>
      <category>Software Architecture</category>
    </item>
    <item>
      <title>Agents Don&apos;t Need Better Models. They Need Better Infrastructure.</title>
      <link>https://twyoon.com/writings/agents-need-better-infrastructure/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/agents-need-better-infrastructure/</guid>
      <pubDate>Tue, 24 Mar 2026 00:00:00 GMT</pubDate>
      <description>300 builders hit inference limits at Mistral&apos;s worldwide hackathon. What agent developers need to know about the serving layer GTC 2026 announced to fix it.</description>
      <category>Agentic AI</category>
      <category>AI Infrastructure</category>
      <category>Open Source</category>
      <category>NVIDIA</category>
      <category>Mistral AI</category>
    </item>
    <item>
      <title>What Senior Business Leaders Actually Ask About AI</title>
      <link>https://twyoon.com/writings/what-senior-business-leaders-ask-about-ai/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/what-senior-business-leaders-ask-about-ai/</guid>
      <pubDate>Tue, 17 Mar 2026 00:00:00 GMT</pubDate>
      <description>Ten questions Directors and C-suite leaders asked when I guest-lectured at Columbia&apos;s executive AI program: liability, data, switching costs, talent.</description>
      <category>Enterprise AI</category>
      <category>AI Strategy</category>
      <category>Executive Education</category>
    </item>
    <item>
      <title>How I Learn Anything with AI</title>
      <link>https://twyoon.com/writings/learning-with-ai/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/learning-with-ai/</guid>
      <pubDate>Thu, 12 Mar 2026 00:00:00 GMT</pubDate>
      <description>How arguing with Claude turned 6 weeks of Causal Inference into a repeatable system for learning any technical subject, with the full chat transcript.</description>
      <category>AI</category>
      <category>Learning</category>
      <category>Claude</category>
      <category>Causal Inference</category>
    </item>
    <item>
      <title>Connecting Local to Global: Inside Mistral AI&apos;s Worldwide Hackathon</title>
      <link>https://twyoon.com/writings/mistral-worldwide-hackathon-2026-recap/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/mistral-worldwide-hackathon-2026-recap/</guid>
      <pubDate>Tue, 03 Mar 2026 00:00:00 GMT</pubDate>
      <description>7,000+ applications, 1,000+ builders, 7 cities, $200K+ in prizes: inside Mistral AI&apos;s first worldwide hackathon, from the judge&apos;s seat at the New York edition.</description>
      <category>Hackathon</category>
      <category>Open Source</category>
      <category>Community Building</category>
      <category>NYC Tech</category>
    </item>
    <item>
      <title>The AI Superpower for PMs: Context Engineering</title>
      <link>https://twyoon.com/writings/ai-superpower-for-pm/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/ai-superpower-for-pm/</guid>
      <pubDate>Thu, 26 Feb 2026 00:00:00 GMT</pubDate>
      <description>Context engineering is the core AI skill for product managers. What it is, how it differs from prompt engineering, and how PMs can practice it.</description>
      <category>Product Management</category>
      <category>Context Engineering</category>
      <category>Claude Code</category>
    </item>
    <item>
      <title>How We Brought 300 Builders to Manhattanville on Valentine&apos;s Day</title>
      <link>https://twyoon.com/writings/iterate-columbia-hackathon-2026-recap/</link>
      <guid isPermaLink="true">https://twyoon.com/writings/iterate-columbia-hackathon-2026-recap/</guid>
      <pubDate>Mon, 16 Feb 2026 00:00:00 GMT</pubDate>
      <description>300 builders, 60+ teams, 8 corporate sponsors, and a 20+ person organizing team. Inside the Valentine&apos;s Day hackathon at Columbia Business School&apos;s Geffen Hall.</description>
      <category>Hackathon</category>
      <category>Agentic AI</category>
      <category>Community Building</category>
      <category>NYC Tech</category>
    </item>
  </channel>
</rss>
