<?xml version="1.0" encoding="UTF-8" ?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>Writing by Aksel Alp</title>
    <link>https://akselalp.com/writing/</link>
    <description>Writing on AI systems, data architecture, product engineering, and systems thinking.</description>
    <language>en</language>
    
    <item>
      <title>Anthropic&apos;s IPO: what it means for the AI industry</title>
      <link>https://akselalp.com/writing/anthropic-ipo/</link>
      <guid isPermaLink="true">https://akselalp.com/writing/anthropic-ipo/</guid>
      <description>What Anthropic&apos;s November 2026 listing disclosed about the economics of frontier AI: compute commitments, enterprise revenue, governance, and risk, and what it suggests for startups and AI engineers.</description>
      
      <dc:creator>Aksel Alp</dc:creator>
      <category>AI industry</category>
<category>AI infrastructure</category>
<category>Analysis</category>
    </item>
<item>
      <title>Autonomous coding agents: Claude Code, Codex, and Grok Build compared</title>
      <link>https://akselalp.com/writing/autonomous-coding-agents/</link>
      <guid isPermaLink="true">https://akselalp.com/writing/autonomous-coding-agents/</guid>
      <description>How today&apos;s coding agents work, what their official documentation says about context, sandboxes, approvals, CI, and review, and the engineering practices that make agent-written code trustworthy.</description>
      
      <dc:creator>Aksel Alp</dc:creator>
      <category>Coding agents</category>
<category>Agents</category>
<category>Software engineering</category>
    </item>
<item>
      <title>Agent orchestration: from individual tools to coordinated AI systems</title>
      <link>https://akselalp.com/writing/agent-orchestration/</link>
      <guid isPermaLink="true">https://akselalp.com/writing/agent-orchestration/</guid>
      <description>What orchestration adds on top of a single agent: routing, delegation, state boundaries, recovery, observability, and approvals, and when a multi-agent design is not worth it.</description>
      
      <dc:creator>Aksel Alp</dc:creator>
      <category>Agents</category>
<category>AI architecture</category>
<category>Reliability</category>
    </item>
<item>
      <title>The agentic loop: how autonomous AI systems actually work</title>
      <link>https://akselalp.com/writing/the-agentic-loop/</link>
      <guid isPermaLink="true">https://akselalp.com/writing/the-agentic-loop/</guid>
      <description>A plain explanation of the observe, reason, act cycle behind AI agents: planning, tool execution, feedback, memory, evaluation, stopping conditions, and why unbounded autonomy is a design flaw.</description>
      
      <dc:creator>Aksel Alp</dc:creator>
      <category>Agents</category>
<category>AI systems</category>
<category>Reliability</category>
    </item>
<item>
      <title>Building reliable AI agents: start with system design</title>
      <link>https://akselalp.com/writing/reliable-ai-agent-systems/</link>
      <guid isPermaLink="true">https://akselalp.com/writing/reliable-ai-agent-systems/</guid>
      <description>A practical framework for moving from a persuasive model demo to an AI agent that can operate inside a real workflow.</description>
      
      <dc:creator>Aksel Alp</dc:creator>
      <category>AI systems</category>
<category>Agents</category>
<category>Reliability</category>
    </item>
<item>
      <title>RAG vs. structured data agents: choose by the question</title>
      <link>https://akselalp.com/writing/rag-vs-structured-data-agents/</link>
      <guid isPermaLink="true">https://akselalp.com/writing/rag-vs-structured-data-agents/</guid>
      <description>Why retrieval over documents and tools over databases solve different information problems, and how to combine them without blurring the boundary.</description>
      
      <dc:creator>Aksel Alp</dc:creator>
      <category>RAG</category>
<category>Data systems</category>
<category>AI architecture</category>
    </item>
  </channel>
</rss>