<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>SubraLabs Lab</title>
    <link>https://subralabs.com/lab/</link>
    <atom:link href="https://subralabs.com/lab/feed.xml" rel="self" type="application/rss+xml" />
    <description>Independent technical analyses from SubraLabs, based on real-world iOS and web development. No client names, no marketing — just data and observations.</description>
    <language>en</language>
    <copyright>Copyright 2026 SubraLabs</copyright>
    <managingEditor>martino@subralabs.com (Martino Vigiani)</managingEditor>
    <webMaster>martino@subralabs.com (Martino Vigiani)</webMaster>
    <lastBuildDate>Thu, 30 Jul 2026 00:00:00 +0000</lastBuildDate>
    <item>
      <title>Fixing DualSense Stick Drift at the Hardware Layer</title>
      <link>https://subralabs.com/lab/sense-calibrator.html</link>
      <guid isPermaLink="true">https://subralabs.com/lab/sense-calibrator.html</guid>
      <pubDate>Sat, 13 Jun 2026 00:00:00 +0000</pubDate>
      <author>martino@subralabs.com (Martino Vigiani)</author>
      <description>Software deadzones only mask controller stick drift on one machine; a WebHID write to the controller's own NVS memory fixes it across PS5, PC and Mac. Method, comparison, and limits from a browser recalibration tool built with Fable 5. Open source.</description>
    </item>
    <item>
      <title>A Decision Framework for Picking the Right LLM in 2026</title>
      <link>https://subralabs.com/lab/llm-chooser.html</link>
      <guid isPermaLink="true">https://subralabs.com/lab/llm-chooser.html</guid>
      <pubDate>Wed, 01 Apr 2026 00:00:00 +0000</pubDate>
      <author>martino@subralabs.com (Martino Vigiani)</author>
      <description>Seven dimensions for choosing a language model when 350+ viable options exist, worked through against the OpenRouter catalog of 353 models from 56 providers.</description>
    </item>
    <item>
      <title>On-Device vs Cloud LLMs for Agentic Tool Calling</title>
      <link>https://subralabs.com/lab/on-device-vs-cloud-llm.html</link>
      <guid isPermaLink="true">https://subralabs.com/lab/on-device-vs-cloud-llm.html</guid>
      <pubDate>Wed, 01 Apr 2026 00:00:00 +0000</pubDate>
      <author>martino@subralabs.com (Martino Vigiani)</author>
      <description>A practical comparison of Apple Foundation Models (~3B) and GPT-OSS 20B for multi-step agentic tool calling in an iOS travel assistant built for a client. Based on real production development.</description>
    </item>
  </channel>
</rss>
