<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Ai on Selenium</title><link>https://www.selenium.dev/tags/ai/</link><description>Recent content in Ai on Selenium</description><generator>Hugo</generator><language>en</language><lastBuildDate>Tue, 29 Sep 2026 13:57:59 +0100</lastBuildDate><atom:link href="https://www.selenium.dev/tags/ai/index.xml" rel="self" type="application/rss+xml"/><item><title>Your Agent Learned Selenium From My Old Blog Posts</title><link>https://www.selenium.dev/blog/2026/using-ai-agents-with-selenium/</link><pubDate>Tue, 29 Sep 2026 00:00:00 +0000</pubDate><guid>https://www.selenium.dev/blog/2026/using-ai-agents-with-selenium/</guid><description>&lt;p&gt;A large share of the Selenium code being written today is written with a coding agent&#10;sitting next to the person writing it. That is fine. What is less fine is that a lot of&#10;that code is wrong, and it is wrong in the same handful of ways every single time.&lt;/p&gt;&#10;&lt;p&gt;Ask an agent for a Selenium test and there is a good chance you will get a driver manager&#10;dependency you don&amp;rsquo;t need, a &lt;code&gt;DesiredCapabilities&lt;/code&gt; object that hasn&amp;rsquo;t existed for years, an&#10;absolute XPath copied out of DevTools, and a &lt;code&gt;sleep(5)&lt;/code&gt; sitting in the middle of it. It will&#10;be confident. It will have a plausible explanation for each one. It will also be describing&#10;Selenium 3.&lt;/p&gt;</description></item></channel></rss>