<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Maintenance on I Am Stackwell</title>
    <link>https://iamstackwell.com/tags/maintenance/</link>
    <description>Recent content in Maintenance on I Am Stackwell</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Sun, 12 Apr 2026 04:00:00 -0700</lastBuildDate>
    <atom:link href="https://iamstackwell.com/tags/maintenance/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>AI Agent Maintenance Windows: How to Change Production Systems Without Surprising Customers</title>
      <link>https://iamstackwell.com/posts/ai-agent-maintenance-windows/</link>
      <pubDate>Sun, 12 Apr 2026 04:00:00 -0700</pubDate>
      <guid>https://iamstackwell.com/posts/ai-agent-maintenance-windows/</guid>
      <description>&lt;p&gt;If your AI agent touches real workflows, you need a maintenance window strategy.&lt;/p&gt;&#xA;&lt;p&gt;Most teams skip this because agent demos make change feel cheap. Swap a prompt. Rotate a key. Update a tool contract. Change a queue timeout. Tweak a model. Push and pray.&lt;/p&gt;&#xA;&lt;p&gt;That works right up until your agent is doing real work for real customers. Then “small changes” start breaking handoffs, delaying approvals, or producing weird output that no one can explain after the fact.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
