<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Spry Techies — Writing</title>
    <link>https://sprytechies.com/blog</link>
    <description>Spry Techies is an AI-native product engineering studio. We design, build, and operate AI products, data systems, and the evals that keep them honest.</description>
    <language>en</language>
    <lastBuildDate>Thu, 12 Nov 2026 00:00:00 GMT</lastBuildDate>
    <generator>Spry Techies static site</generator>
    <atom:link href="https://sprytechies.com/rss.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Eval-driven delivery is the real AI-SDLC shift</title>
      <link>https://sprytechies.com/blog/eval-driven-delivery</link>
      <guid isPermaLink="true">https://sprytechies.com/blog/eval-driven-delivery</guid>
      <pubDate>Thu, 12 Nov 2026 00:00:00 GMT</pubDate>
      <description>AI software changes on every prompt and model update. Evaluation is how a team keeps that change from becoming drift.</description>
      <category>ai-sdlc</category>
      <category>evaluation</category>
    </item>
    <item>
      <title>Spec-driven development is the AI-SDLC control plane</title>
      <link>https://sprytechies.com/blog/spec-driven-development-as-ai-sdlc</link>
      <guid isPermaLink="true">https://sprytechies.com/blog/spec-driven-development-as-ai-sdlc</guid>
      <pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate>
      <description>Evals catch regressions. A spec defines what the system is supposed to do. AI-SDLC needs both, linked at the clause level.</description>
      <category>ai-sdlc</category>
      <category>spec-driven</category>
    </item>
    <item>
      <title>Guardrails that belong in the product, not the prompt</title>
      <link>https://sprytechies.com/blog/guardrails-in-the-product</link>
      <guid isPermaLink="true">https://sprytechies.com/blog/guardrails-in-the-product</guid>
      <pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate>
      <description>A prompt instruction is a suggestion. Production guardrails are code, permissions, and review paths the model cannot talk its way around.</description>
      <category>ai-engineering</category>
      <category>guardrails</category>
    </item>
  </channel>
</rss>
