<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>AI-Enabled Software on Chetan Dalal — Senior Software Engineer</title><link>https://chetandalal.net/categories/ai-enabled-software/</link><description>Recent content in AI-Enabled Software on Chetan Dalal — Senior Software Engineer</description><generator>Hugo -- gohugo.io</generator><language>en-US</language><copyright>© 2019–2026 Chetan Dalal</copyright><lastBuildDate>Sat, 01 Aug 2026 01:25:29 -0400</lastBuildDate><atom:link href="https://chetandalal.net/categories/ai-enabled-software/index.xml" rel="self" type="application/rss+xml"/><item><title>Evaluating an AI Feature When There's No Ground Truth</title><link>https://chetandalal.net/posts/evaluating-ai-without-ground-truth/</link><pubDate>Sat, 01 Aug 2026 00:00:00 +0000</pubDate><guid>https://chetandalal.net/posts/evaluating-ai-without-ground-truth/</guid><description>Most AI features ship without a defensible way to tell whether a change made them better. Here&amp;rsquo;s how to build an evaluation approach when ground truth doesn&amp;rsquo;t exist, and how to be honest about what it can&amp;rsquo;t tell you.</description></item><item><title>Intelligent Workflow Platform</title><link>https://chetandalal.net/projects/intelligent-workflow-platform/</link><pubDate>Fri, 31 Jul 2026 00:00:00 +0000</pubDate><guid>https://chetandalal.net/projects/intelligent-workflow-platform/</guid><description>Automating document-heavy workflows where a wrong answer is expensive, and why the human review step turned out to be the part that determined whether the system was worth running.</description></item></channel></rss>