<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Pranshul Chandhok | Notes</title><description>Field notes on production AI, real-time 3D, browser-native systems, and product reliability.</description><link>https://www.2600th.com/</link><language>en-IN</language><item><title>The hard part of AI floorplan parsing isn’t the model</title><link>https://www.2600th.com/notes/ai-floorplan-parsing/</link><guid isPermaLink="true">https://www.2600th.com/notes/ai-floorplan-parsing/</guid><description>A source-grounded look at a floorplan pipeline: where structured output helps, how file-based resume can fail, and which validation mechanisms still belong in the next version.</description><pubDate>Wed, 02 Sep 2026 00:00:00 GMT</pubDate><category>Applied AI</category><category>Spatial systems</category><category>Pipeline reliability</category><author>Pranshul Chandhok</author></item><item><title>The boundary between generative and deterministic systems</title><link>https://www.2600th.com/notes/generative-and-deterministic-systems/</link><guid isPermaLink="true">https://www.2600th.com/notes/generative-and-deterministic-systems/</guid><description>Designesto, Blocks and the PropVR AI foundation show three different places to draw the boundary between open-ended generation and decisions a product must preserve.</description><pubDate>Wed, 02 Sep 2026 00:00:00 GMT</pubDate><category>Product architecture</category><category>Applied AI</category><category>Design technology</category><author>Pranshul Chandhok</author></item><item><title>AI video got good. Directing a sequence is still hard.</title><link>https://www.2600th.com/notes/ai-video-control/</link><guid isPermaLink="true">https://www.2600th.com/notes/ai-video-control/</guid><description>A thirty-second experiment, two saved local H3 control graphs, and a proposed six-beat sequence show why shot quality and editorial continuity are different problems.</description><pubDate>Mon, 24 Aug 2026 00:00:00 GMT</pubDate><category>AI video</category><category>Creative direction</category><category>Local generation</category><author>Pranshul Chandhok</author></item><item><title>Making tools easier to steer</title><link>https://www.2600th.com/notes/technology-and-human-agency/</link><guid isPermaLink="true">https://www.2600th.com/notes/technology-and-human-agency/</guid><description>In design software and AI tools, I keep coming back to a practical question. Can someone change the result without having to start over?</description><pubDate>Sun, 23 Aug 2026 00:00:00 GMT</pubDate><category>Human agency</category><category>Product design</category><category>Immersive systems</category><author>Pranshul Chandhok</author></item><item><title>What I’m revisiting in my 2023 essay on AI and games</title><link>https://www.2600th.com/notes/ai-native-game-development-reflection/</link><guid isPermaLink="true">https://www.2600th.com/notes/ai-native-game-development-reflection/</guid><description>I wrote about AI in game development in 2023. Building with these tools has since made me more interested in revision, consistency, and what happens inside the editor.</description><pubDate>Sat, 22 Aug 2026 00:00:00 GMT</pubDate><category>Applied AI</category><category>Game development</category><category>Production systems</category><author>Pranshul Chandhok</author></item><item><title>Why I still build things myself</title><link>https://www.2600th.com/notes/from-pixels-to-intelligent-systems/</link><guid isPermaLink="true">https://www.2600th.com/notes/from-pixels-to-intelligent-systems/</guid><description>I’ve moved from game programming to leading product and technology teams. Making things myself still helps me understand what to ask of a product and the people building it.</description><pubDate>Fri, 21 Aug 2026 00:00:00 GMT</pubDate><category>Career reflections</category><category>Product engineering</category><category>Emerging technology</category><author>Pranshul Chandhok</author></item><item><title>From a documentary to a browser flight experiment</title><link>https://www.2600th.com/notes/browser-flight-experiment/</link><guid isPermaLink="true">https://www.2600th.com/notes/browser-flight-experiment/</guid><description>After watching Operation Safed Sagar, I spent a few days building a fictional reconnaissance flight that runs in the browser.</description><pubDate>Sat, 15 Aug 2026 00:00:00 GMT</pubDate><category>Browser games</category><category>AI-assisted engineering</category><category>Real-time 3D</category><author>Pranshul Chandhok</author></item><item><title>Web Ocean 3D: what broke when other people ran it</title><link>https://www.2600th.com/notes/ocean-reliability/</link><guid isPermaLink="true">https://www.2600th.com/notes/ocean-reliability/</guid><description>Sharing my browser ocean exposed a texture budget I was not measuring. The repair cut asset allocation, increased download weight, and revealed a broken visual baseline.</description><pubDate>Sun, 09 Aug 2026 00:00:00 GMT</pubDate><category>Reliability</category><category>Web graphics</category><category>Performance</category><author>Pranshul Chandhok</author></item></channel></rss>