<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/"><channel><title>Aixy Engineering</title><link>https://aixy-gateway.com/blog/</link><description>How we build, what we measure, and what we learn at Aixy.</description><language>en</language><atom:link href="https://aixy-gateway.com/rss.xml" rel="self" type="application/rss+xml"/><item><title>Decision models on Aixy: from probabilities to application actions</title><link>https://aixy-gateway.com/blog/decision-models-on-aixy/</link><guid isPermaLink="true">https://aixy-gateway.com/blog/decision-models-on-aixy/</guid><description>An AI assistant needs to decide when to answer, search or ask for review. Learn how Jev, Clef and GLiDE differ, why confidence thresholds do not transfer automatically, and how to evaluate a decision workflow through Aixy.</description><pubDate>Mon, 19 Oct 2026 13:49:34 GMT</pubDate><category>Product</category><dc:creator>Óscar Caballero</dc:creator></item><item><title>When MCP tools fill the context before the work begins</title><link>https://aixy-gateway.com/blog/mcp-tool-discovery-context/</link><guid isPermaLink="true">https://aixy-gateway.com/blog/mcp-tool-discovery-context/</guid><description>Growing MCP catalogs consume model context before work begins. We examine discovery, encoding and the measurements from our experiments, including what they leave unresolved.</description><pubDate>Wed, 07 Oct 2026 05:00:00 GMT</pubDate><category>MCP</category><dc:creator>Óscar Caballero</dc:creator></item></channel></rss>