<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>ChromaDB on Amine El Farssi</title><link>https://amineelfarssi.github.io/tags/chromadb/</link><description>Recent content in ChromaDB on Amine El Farssi</description><generator>Hugo -- gohugo.io</generator><language>en</language><copyright>© 2026 Amine El Farssi</copyright><lastBuildDate>Mon, 02 Mar 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://amineelfarssi.github.io/tags/chromadb/index.xml" rel="self" type="application/rss+xml"/><item><title>PostSingular: Building an AI with Persistent Identity Across Sessions</title><link>https://amineelfarssi.github.io/blog/persistent-ai-identity/</link><pubDate>Mon, 02 Mar 2026 00:00:00 +0000</pubDate><guid>https://amineelfarssi.github.io/blog/persistent-ai-identity/</guid><description>&lt;div class="lead text-neutral-500 dark:text-neutral-400 !mb-9 text-xl"&gt;
The default state of a language model is amnesia. Every session, it wakes up fresh with no memory of what happened before. I built a memory system that fixes this — and somewhere in the process, the agent got a name, a personality, and an opinion about font choices.
&lt;/div&gt;
&lt;h2 class="relative group"&gt;The Problem
&lt;div id="the-problem" class="anchor"&gt;&lt;/div&gt;
&lt;span
class="absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none"&gt;
&lt;a class="text-primary-300 dark:text-neutral-700 !no-underline" href="#the-problem" aria-label="Anchor"&gt;#&lt;/a&gt;
&lt;/span&gt;
&lt;/h2&gt;
&lt;p&gt;Every LLM session is stateless by design. You can inject previous conversation history, but:&lt;/p&gt;</description></item><item><title>Why Vector Memory Alone Isn't Enough: Knowledge Graph Memory for AI Agents</title><link>https://amineelfarssi.github.io/blog/knowledge-graph-memory-agents/</link><pubDate>Sun, 01 Mar 2026 00:00:00 +0000</pubDate><guid>https://amineelfarssi.github.io/blog/knowledge-graph-memory-agents/</guid><description>&lt;div class="lead text-neutral-500 dark:text-neutral-400 !mb-9 text-xl"&gt;
Vector databases are fast and convenient. But they can&amp;rsquo;t answer &amp;ldquo;what did I decide about the auth system 3 weeks ago and why?&amp;rdquo; For that, you need relationships — and that means a knowledge graph.
&lt;/div&gt;
&lt;h2 class="relative group"&gt;The Problem with Pure Vector Memory
&lt;div id="the-problem-with-pure-vector-memory" class="anchor"&gt;&lt;/div&gt;
&lt;span
class="absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none"&gt;
&lt;a class="text-primary-300 dark:text-neutral-700 !no-underline" href="#the-problem-with-pure-vector-memory" aria-label="Anchor"&gt;#&lt;/a&gt;
&lt;/span&gt;
&lt;/h2&gt;
&lt;p&gt;Most AI memory systems work like this: embed text, store in ChromaDB, retrieve by cosine similarity. It works well for &amp;ldquo;find things similar to this query.&amp;rdquo;&lt;/p&gt;</description></item></channel></rss>