<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>netogrit — Writing</title><description>Essays on evals, cost models and how we work, by the people who do the work.</description><link>https://netogrit.com/</link><language>en-gb</language><atom:link href="https://netogrit.com/rss.xml" rel="self" type="application/rss+xml"/><item><title>A cost model is a deliverable</title><link>https://netogrit.com/writing/cost-model-is-a-deliverable/</link><guid isPermaLink="true">https://netogrit.com/writing/cost-model-is-a-deliverable/</guid><description>The first question about any production AI system should be the cost per request at volume. Most teams find out from the first full invoice.</description><pubDate>Thu, 30 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Why we publish the evals with the model</title><link>https://netogrit.com/writing/publish-the-evals/</link><guid isPermaLink="true">https://netogrit.com/writing/publish-the-evals/</guid><description>A model without its evaluation set is a demo. The eval set is the contract: what the model is for, what it is not, and how you would know it broke.</description><pubDate>Thu, 18 Jun 2026 00:00:00 GMT</pubDate></item><item><title>The pyramid is the problem, not the people</title><link>https://netogrit.com/writing/pyramid-is-the-problem/</link><guid isPermaLink="true">https://netogrit.com/writing/pyramid-is-the-problem/</guid><description>Most failed engagements were staffed by capable people. They failed because the person who understood the problem was not the person who built it.</description><pubDate>Sat, 02 May 2026 00:00:00 GMT</pubDate></item></channel></rss>