<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>cdrift Field Notes</title><description>Decisions, failures, and constraints behind software built and operated at cdrift.</description><link>https://cdrift.com/</link><language>en-us</language><item><title>Retiring Three Frontends Without Rewriting the Product</title><link>https://cdrift.com/notes/retiring-three-frontends/</link><guid isPermaLink="true">https://cdrift.com/notes/retiring-three-frontends/</guid><description>Convergence is a product and operations decision, not a contest to select the most interesting framework.</description><pubDate>Sun, 23 Aug 2026 00:00:00 GMT</pubDate><category>product engineering</category><category>migration</category></item><item><title>Make the AI Pipeline Resumable Before Making It Smarter</title><link>https://cdrift.com/notes/resumable-ai-pipelines/</link><guid isPermaLink="true">https://cdrift.com/notes/resumable-ai-pipelines/</guid><description>Reliability and cost controls matter more than another model call when research runs span unreliable systems.</description><pubDate>Sat, 22 Aug 2026 00:00:00 GMT</pubDate><category>applied AI</category><category>pipelines</category></item><item><title>When Two Backends Must Speak the Same Real-Time Protocol</title><link>https://cdrift.com/notes/two-backends-one-protocol/</link><guid isPermaLink="true">https://cdrift.com/notes/two-backends-one-protocol/</guid><description>What it takes to make a backend fallback operational rather than architectural theater.</description><pubDate>Fri, 21 Aug 2026 00:00:00 GMT</pubDate><category>real-time systems</category><category>reliability</category></item></channel></rss>