<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel><title>Eshaan Mathakari: writing</title><link>index.html</link><description>AI engineer, data scientist, and the person running a small lab out of his apartment.</description><language>en</language><item><title>What an ant colony taught us about teams of AI models</title><link>writing/colony-experiment.html</link><guid>writing/colony-experiment.html</guid><pubDate>Tue, 27 Oct 2026 12:00:00 +0000</pubDate><description>We put several different language models into one simulated colony to see if the mix would beat its best member. The answer pointed somewhere more useful.</description></item><item><title>Six months of agents in my apartment</title><link>writing/six-months-of-agents.html</link><guid>writing/six-months-of-agents.html</guid><pubDate>Tue, 27 Oct 2026 12:00:00 +0000</pubDate><description>I&#x27;ve run a fleet of AI agents on a Mac Studio at home since spring. What held up, what didn&#x27;t, and the failures that looked exactly like success.</description></item><item><title>About ten dollars</title><link>writing/about-ten-dollars.html</link><guid>writing/about-ten-dollars.html</guid><pubDate>Tue, 27 Oct 2026 12:00:00 +0000</pubDate><description>Months of building prediction-market bots earned roughly the price of a sandwich. Here is what the money couldn&#x27;t measure.</description></item></channel></rss>
