<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>AtomsAI blog</title><description>Short pieces on what breaks between a demo and production, written by the engineers who fixed it.</description><link>https://atomsai.com</link><language>en</language><item><title>The demo was never the hard part</title><link>https://atomsai.com/resources/blog/the-demo-was-never-the-hard-part</link><guid isPermaLink="true">https://atomsai.com/resources/blog/the-demo-was-never-the-hard-part</guid><description>Pilots don&apos;t fail on model quality. They fail on the twelve systems between the demo and a finished piece of work — and on the exception queue nobody scoped.</description><pubDate>Wed, 19 Aug 2026 00:00:00 GMT</pubDate><category>production-ai</category><category>integration</category><category>enterprise</category></item><item><title>Write the evals before the prompt</title><link>https://atomsai.com/resources/blog/write-the-evals-before-the-prompt</link><guid isPermaLink="true">https://atomsai.com/resources/blog/write-the-evals-before-the-prompt</guid><description>An eval suite isn&apos;t QA you add at the end. It&apos;s the specification — and the only thing that lets a CXO answer &apos;is it working?&apos; with a number instead of an opinion.</description><pubDate>Wed, 08 Jul 2026 00:00:00 GMT</pubDate><category>evals</category><category>reliability</category><category>governance</category></item><item><title>How much autonomy should an agent have?</title><link>https://atomsai.com/resources/blog/how-much-autonomy-should-an-agent-have</link><guid isPermaLink="true">https://atomsai.com/resources/blog/how-much-autonomy-should-an-agent-have</guid><description>The right answer is a configuration value, not a philosophy. Design the approval gates before the tools, start tight, and let the pass rate buy each loosening.</description><pubDate>Wed, 27 May 2026 00:00:00 GMT</pubDate><category>agents</category><category>governance</category><category>risk</category></item></channel></rss>