<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>Jevgeni Agnevstsikov — Writing</title>
  <link>https://jevgenitig.github.io/my-apps/writing/</link>
  <atom:link href="https://jevgenitig.github.io/my-apps/feed.xml" rel="self" type="application/rss+xml"/>
  <description>Notes from building small Apple-platform apps: Swift, on-device machine
    learning, App Store practice.</description>
  <language>en</language>
  <item>
    <title>Foundation Models against a rule-based parser</title>
    <link>https://jevgenitig.github.io/my-apps/writing/on-device-llm-vs-rules.html</link>
    <guid isPermaLink="true">https://jevgenitig.github.io/my-apps/writing/on-device-llm-vs-rules.html</guid>
    <description>Apple's on-device model versus three hundred lines of rules, on wine labels recognised by Vision: the API's sharp edges, a reproducible benchmark, and why the rules stayed.</description>
    <pubDate>Thu, 24 Sep 2026 09:00:00 +0000</pubDate>
  </item>
</channel>
</rss>
