<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Architecture on NotNil Blog</title><link>https://notnil.org/en/tags/architecture/</link><description>Recent content in Architecture on NotNil Blog</description><generator>Hugo</generator><language>en-US</language><copyright>© Matthieu Moreau</copyright><lastBuildDate>Wed, 05 Feb 2025 00:00:00 +0000</lastBuildDate><atom:link href="https://notnil.org/en/tags/architecture/index.xml" rel="self" type="application/rss+xml"/><item><title>Our GenAI tech stack in production</title><link>https://notnil.org/en/posts/24-squid/</link><pubDate>Wed, 05 Feb 2025 00:00:00 +0000</pubDate><guid>https://notnil.org/en/posts/24-squid/</guid><description>&lt;figure&gt;&lt;img src="https://notnil.org/images/posts/squid/feature.png"&gt;&#10;&lt;/figure&gt;&#10;&#10;&lt;p&gt;Let&amp;rsquo;s talk GenAI tech (but not LLMs)!&lt;/p&gt;&#10;&lt;p&gt;I was sharing &lt;a href="https://notnil.org/en/posts/23-rex2024/"&gt;my Monday observations&lt;/a&gt; on the GenAI market, and notably the importance of a robust software foundation.&lt;/p&gt;&#10;&lt;h2 id="the-fundamentals-of-a-genai-platform"&gt;The fundamentals of a GenAI platform&lt;/h2&gt;&#10;&lt;p&gt;A GenAI platform rests on fundamental elements: task orchestration (often asynchronous), agents with memory management, intelligent document parsing, a high-performance vector database, and fine-grained observability of LLM generation.&lt;/p&gt;&#10;&lt;p&gt;If you do a Google search, sorry, Perplexity, you&amp;rsquo;ll be offered LangChain or llamaIndex, 2 &amp;ldquo;popular&amp;rdquo; frameworks that have the elements cited above. In the end we set LangChain aside; its abstraction layer turned out to be more of a hindrance than anything else. A GenAI project is above all a project of flow orchestration and API calls, nothing that requires a framework.&lt;/p&gt;</description></item></channel></rss>