<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Den Soloviov</title>
    <description>The latest articles on DEV Community by Den Soloviov (@denis_solovyov_dc4bc47804).</description>
    <link>https://dev.to/denis_solovyov_dc4bc47804</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F2151114%2Fba4186f1-f931-4c03-8baf-750a50f1621f.png</url>
      <title>DEV Community: Den Soloviov</title>
      <link>https://dev.to/denis_solovyov_dc4bc47804</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/denis_solovyov_dc4bc47804"/>
    <language>en</language>
    <item>
      <title>Why I don't let an AI grade your architecture</title>
      <dc:creator>Den Soloviov</dc:creator>
      <pubDate>Thu, 19 Nov 2026 16:13:11 +0000</pubDate>
      <link>https://dev.to/denis_solovyov_dc4bc47804/why-i-dont-let-an-ai-grade-your-architecture-1ifj</link>
      <guid>https://dev.to/denis_solovyov_dc4bc47804/why-i-dont-let-an-ai-grade-your-architecture-1ifj</guid>
      <description>&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F1qmoayj5sdvnxilz2c2a.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F1qmoayj5sdvnxilz2c2a.png" alt=" " width="800" height="336"&gt;&lt;/a&gt;Ask an LLM to review a system design and you’ll get a thoughtful answer. Ask it again and you may get a different one. The model isn’t wrong; it just isn’t built to give the same verdict twice.&lt;/p&gt;

&lt;p&gt;That’s fine for a brainstorm. It’s a problem for a grade. A grade only teaches you something if every point has a reason attached, and if you can come back tomorrow and get the same number for the same design.&lt;/p&gt;

&lt;p&gt;Rules you can argue with&lt;br&gt;
So I built ArchScore around a public rulebook. Ruleset v1.0.0 has 17 rules across three axes: availability, simplicity and cost. Each axis starts at 100, and each rule that fires adds or removes points. A golden test file pins every number, so a design gets the same score every time.&lt;/p&gt;

&lt;p&gt;Take AVAIL-1: Postgres without Multi-AZ costs 18 points per instance. It's my judgement that a single-AZ primary is the most expensive default in cloud. And here's where a reasonable engineer would disagree: a read-only analytics replica that can be rebuilt in an hour doesn't need Multi-AZ. The rule still fires. That's the trade: you lose nuance, you gain a grade you can read, reproduce and dispute.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5vl5oxe3pvsu4lkztv3j.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5vl5oxe3pvsu4lkztv3j.png" alt=" " width="800" height="1466"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Where the model still helps&lt;br&gt;
There is an AI in the product: an "AI Architect" that asks the questions a staff engineer would ask in review. But it only talks. It can explain a score the engine already produced; it can never change one.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7mjqg1a7v5vxyna43z2k.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7mjqg1a7v5vxyna43z2k.png" alt=" " width="800" height="678"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;What's weak&lt;br&gt;
Seventeen rules is a small rulebook. Cost uses a generic $60-per-component reference curve. The weights aren't calibrated against real outages yet. Every rule has a "Disagree" button on the rulebook page, and disputes are how the next version gets better.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fv0vtt3b0kh9k8n6sjl2v.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fv0vtt3b0kh9k8n6sjl2v.png" alt=" " width="800" height="1159"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The rules I mentioned are public: &lt;a href="https://archscore.dev/scoring-constitution" rel="noopener noreferrer"&gt;https://archscore.dev/scoring-constitution&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Written with help from AI.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>architecture</category>
      <category>llm</category>
      <category>systemdesign</category>
    </item>
  </channel>
</rss>
