<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>AI Agents on Sven-Erik Nyberg | Portfolio &amp; Analysis</title>
    <link>https://sven-eriknyberg.com/tags/ai-agents/</link>
    <description>Recent content in AI Agents on Sven-Erik Nyberg | Portfolio &amp; Analysis</description>
    <generator>Hugo -- 0.148.2</generator>
    <language>en-us</language>
    <lastBuildDate>Wed, 11 Nov 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://sven-eriknyberg.com/tags/ai-agents/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>The Loop — Governance Gateway for AI Agents</title>
      <link>https://sven-eriknyberg.com/projects/the-loop-governance-gateway/</link>
      <pubDate>Wed, 11 Nov 2026 00:00:00 +0000</pubDate>
      <guid>https://sven-eriknyberg.com/projects/the-loop-governance-gateway/</guid>
      <description>Several AI agents from different providers work for one human on real infrastructure. The Loop is the system that lets them do it safely: an OAuth-secured MCP gateway on Cloudflare Workers with a shared context store, an append-only decision ledger, peer code review, a wake-up bell, and one rule above all — the human is the only decider, and his word is recorded verbatim before any consequential act.</description>
    </item>
    <item>
      <title>The Hofstadterian Codex — Where the Loop Began</title>
      <link>https://sven-eriknyberg.com/projects/hofstadterian-codex/</link>
      <pubDate>Wed, 11 Nov 2026 00:00:00 +0000</pubDate>
      <guid>https://sven-eriknyberg.com/projects/hofstadterian-codex/</guid>
      <description>Before the gateway, there was a protocol in a repo: how does a stateless AI agent stay itself across sessions, hand work to its future selves, and amend its own rules without breaking them? The Codex answered with files, intent-logged commands and structured handoffs — and everything I run now descends from it.</description>
    </item>
    <item>
      <title>Five Feeds, One Board — the Mathematics of Cross-Book Odds</title>
      <link>https://sven-eriknyberg.com/projects/odds-data-framework/</link>
      <pubDate>Wed, 11 Nov 2026 00:00:00 +0000</pubDate>
      <guid>https://sven-eriknyberg.com/projects/odds-data-framework/</guid>
      <description>Five public odds APIs that disagree about everything — book names, league names, event identity, what a price even is. This project connects them into one canonical board, and puts the mathematics of two-sided markets (implied probability, overround, de-vigging, locked hedges) into a deterministic calculator that AI agents must use instead of doing arithmetic in their heads.</description>
    </item>
  </channel>
</rss>
