<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
	<channel>
		<title>Evals on Blog by Thiyagaraj T</title>
		<link>https://thiyaga.xyz/tags/evals/</link>
		<description>Recent content in Evals on Blog by Thiyagaraj T</description>
		<generator>Hugo</generator>
		<language>en</language>
		
		
		
		
			<lastBuildDate>Wed, 28 Oct 2026 00:30:00 +0530</lastBuildDate>
		
			<atom:link href="https://thiyaga.xyz/tags/evals/index.xml" rel="self" type="application/rss+xml" />
			<item>
				<title>Evaluating Conversation Agents</title>
				<link>https://thiyaga.xyz/posts/conversation-quality-metrics/</link>
				<pubDate>Wed, 28 Oct 2026 00:30:00 +0530</pubDate>
				<guid>https://thiyaga.xyz/posts/conversation-quality-metrics/</guid>
				<description>&lt;p&gt;&lt;em&gt;Say you run an agent that does 20-minute phone screens. It greets the candidate, asks about their last two roles, digs into one project, answers questions about the job, and wraps up. Thousands of calls a day. You change one line in the prompt. Did the calls get better or worse?&lt;/em&gt;&lt;/p&gt;&#xA;&lt;p&gt;&lt;em&gt;This post is how we answer that question.&lt;/em&gt;&lt;/p&gt;&#xA;&lt;hr&gt;&#xA;&lt;h2 id=&#34;1-what-does-not-work&#34;&gt;1. What does not work&lt;/h2&gt;&#xA;&lt;p&gt;&lt;img src=&#34;https://thiyaga.xyz/images/cq-no-golden.svg&#34; alt=&#34;A task eval diffs one output against one reference; a conversation has three good next questions and nothing to diff against&#34;&gt;&lt;/p&gt;</description>
			</item>
	</channel>
</rss>
