<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/dd586dc734294914ba4d4e2c16610161&quot; frameborder=&quot;0&quot; width=&quot;1832&quot; height=&quot;1374&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1374</height><width>1832</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1374</thumbnail_height><thumbnail_width>1832</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/dd586dc734294914ba4d4e2c16610161-e7eb8a86f66473bd.gif</thumbnail_url><duration>472.638</duration><title>SWE - Benchmark - AI Agents - Update</title><description>In this video, I discuss the groundbreaking advancements in AI software engineering, particularly focusing on the impressive performance of Devon AI on the SWE benchmark. I highlight how Honeycomb has excelled in resolving GitHub issues and the significant impact of Amazon Q in saving development time and costs. Additionally, I touch upon the unique approach of ChattyPT in problem-solving and the innovative work of Sakana AI in producing cutting-edge scientific papers. Viewers are encouraged to explore these AI technologies and their potential applications.</description></oembed>