<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Notes from the work · Esya</title><description>Field notes on taking AI workflows into production.</description><link>https://esya.studio</link><language>en-GB</language><item><title>Evals, runbook and handover: what must ship with an AI workflow</title><link>https://esya.studio/notes/ai-evals-runbook-handover</link><guid isPermaLink="true">https://esya.studio/notes/ai-evals-runbook-handover</guid><description>The three operating artefacts belong in one chain: evidence informs release, production signals lead to action, and the receiving team can run the next change.</description><pubDate>Sat, 29 Aug 2026 00:00:00 GMT</pubDate><category>production-ai</category></item><item><title>How to choose the first AI workflow worth building</title><link>https://esya.studio/notes/choose-first-ai-workflow</link><guid isPermaLink="true">https://esya.studio/notes/choose-first-ai-workflow</guid><description>A written screen for choosing one consequential workflow without a fake ROI score, an unbounded use-case list or a model looking for a job.</description><pubDate>Sat, 29 Aug 2026 00:00:00 GMT</pubDate><category>production-ai</category></item><item><title>What production actually means for an AI workflow</title><link>https://esya.studio/notes/production-ai-workflow</link><guid isPermaLink="true">https://esya.studio/notes/production-ai-workflow</guid><description>A five-part test for the point where an AI workflow stops being a demonstration and becomes owned, observable work inside real systems.</description><pubDate>Sat, 29 Aug 2026 00:00:00 GMT</pubDate><category>production-ai</category></item></channel></rss>