<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Copilot-Studio on Pavel Nasovich's Blog</title><link>https://forcewake.me/tags/copilot-studio/</link><description>Recent content in Copilot-Studio on Pavel Nasovich's Blog</description><generator>Hugo -- 0.157.0</generator><language>en-us</language><copyright>Copyright 2026</copyright><lastBuildDate>Sat, 02 Aug 2025 18:39:41 +0200</lastBuildDate><atom:link href="https://forcewake.me/tags/copilot-studio/index.xml" rel="self" type="application/rss+xml"/><item><title>How I Stopped My Copilot Studio Agent From Hallucinating (And You Can Too)</title><link>https://forcewake.me/how-i-stopped-my-copilot-studio-agent-from-hallucinating-and-you-can-too/</link><pubDate>Sat, 02 Aug 2025 00:00:00 +0000</pubDate><guid>https://forcewake.me/how-i-stopped-my-copilot-studio-agent-from-hallucinating-and-you-can-too/</guid><description>After a month of battling hallucinating AI agents and midnight support calls about features that don&amp;#39;t exist, here&amp;#39;s the raw truth about building reliable Copilot Studio agents. This deep dive reveals why that innocent &amp;#34;general knowledge&amp;#34; toggle is your worst enemy, how to write paranoid prompts that actually work, and why teaching your agent to say &amp;#34;I don&amp;#39;t know&amp;#34; might be the best feature you&amp;#39;ll ever implement. We share our painful journey from creative writing bot to trustworthy assistant—including all the facepalm moments Microsoft won&amp;#39;t tell you about.</description></item></channel></rss>