<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Jerem Flow — Updates</title>
  <subtitle>Blog and notes — analytical chemistry PhD, mass spectrometry, lab informatics and AI.</subtitle>
  <link href="https://jeremflow.com/feed/feed.xml" rel="self" />
  <link href="https://jeremflow.com/" />
  <updated>2026-08-21T22:31:00.000Z</updated>
  <id>https://jeremflow.com/</id>
  <author><name>Jérémy</name></author>
  
  <entry>
    <title>Connecting Agilent mass-spec data to an AI — a first mapping</title>
    <link href="https://jeremflow.com/projects/connecting-agilent-ms-data-to-ai/" />
    <updated>2026-08-21T22:31:00.000Z</updated>
    <id>https://jeremflow.com/projects/connecting-agilent-ms-data-to-ai/</id>
    <summary>A personal exploration of how to connect Agilent LC-MS data to AI for compound annotation — the state of the art, how MassHunter Explorer 2.0 fits in, what&#39;s available as open-source engines, and the concrete path I&#39;m sketching for a pilot.</summary>
  </entry>
  
  <entry>
    <title>Why I moved my notes from Notion to Obsidian</title>
    <link href="https://jeremflow.com/notes/notion-to-obsidian/" />
    <updated>2026-08-21T15:30:00.000Z</updated>
    <id>https://jeremflow.com/notes/notion-to-obsidian/</id>
    <summary>The real reasons I left Notion for Obsidian — privacy, a RAG that actually searches my notes, automation and AI interaction, and a cost that beats Notion Business while giving me more.</summary>
  </entry>
  
  <entry>
    <title>Using a reMarkable Paper Pro without the subscription</title>
    <link href="https://jeremflow.com/notes/remarkable-without-subscription/" />
    <updated>2026-08-19T12:10:00.000Z</updated>
    <id>https://jeremflow.com/notes/remarkable-without-subscription/</id>
    <summary>Everything the reMarkable Connect subscription gates, and the free workarounds I use instead — backup, handwriting-to-text, cloud sync, and e-book transfer.</summary>
  </entry>
  
  <entry>
    <title>Turning my e-book library into a searchable, private knowledge base</title>
    <link href="https://jeremflow.com/projects/base-livres/" />
    <updated>2026-08-15T16:25:00.000Z</updated>
    <id>https://jeremflow.com/projects/base-livres/</id>
    <summary>How I digest my e-books into Markdown, index them into a private RAG, and write rich Obsidian summaries — so the ideas in my books become searchable and reusable.</summary>
  </entry>
  
  <entry>
    <title>A data-driven personal coach, built from my own health data</title>
    <link href="https://jeremflow.com/projects/coach-sportif-ia/" />
    <updated>2026-08-10T07:30:00.000Z</updated>
    <id>https://jeremflow.com/projects/coach-sportif-ia/</id>
    <summary>How I turned my own training, nutrition and sleep data into a private, data-driven coach — the sources, the DuckDB pipeline, the dashboard, and the agent that answers with real numbers.</summary>
  </entry>
  
  <entry>
    <title>Camoufox vs browserless — evading bot detection the honest way</title>
    <link href="https://jeremflow.com/notes/camoufox-vs-browserless/" />
    <updated>2026-08-03T21:15:00.000Z</updated>
    <id>https://jeremflow.com/notes/camoufox-vs-browserless/</id>
    <summary>The difference between a stock headless browser and a stealth one, why browserless gets detected, and how Camoufox handles hardened sites — and the line I don&#39;t cross.</summary>
  </entry>
  
  <entry>
    <title>The guardrails of an autonomous agent</title>
    <link href="https://jeremflow.com/notes/guardrails-autonomous-agent/" />
    <updated>2026-07-22T17:40:00.000Z</updated>
    <id>https://jeremflow.com/notes/guardrails-autonomous-agent/</id>
    <summary>The safety and reliability patterns I put around my autonomous agent — watchdogs, scheduled jobs, alerts, and self-healing — so it can run unattended without breaking silently.</summary>
  </entry>
  
  <entry>
    <title>Giving Hermes a headless browser (and a VPN) for the web</title>
    <link href="https://jeremflow.com/projects/headless-browser-hermes/" />
    <updated>2026-07-10T08:50:00.000Z</updated>
    <id>https://jeremflow.com/projects/headless-browser-hermes/</id>
    <summary>Why and how I gave my Hermes agent a real headless browser — a self-hosted Chromium it can drive over CDP, with all traffic routed through a VPN. What it unlocks, how it&#39;s wired up, and the guardrails I put in place.</summary>
  </entry>
  
  <entry>
    <title>Why I self-host my AI assistant</title>
    <link href="https://jeremflow.com/notes/self-hosting-ai-assistant/" />
    <updated>2026-06-28T19:05:00.000Z</updated>
    <id>https://jeremflow.com/notes/self-hosting-ai-assistant/</id>
    <summary>The real reasons I run my AI assistant on my own hardware instead of a SaaS — privacy, cost, control, and what it actually costs me.</summary>
  </entry>
  
  <entry>
    <title>Hindsight + Hermes — giving my assistant a real long-term memory</title>
    <link href="https://jeremflow.com/projects/hindsight-hermes/" />
    <updated>2026-06-15T10:20:00.000Z</updated>
    <id>https://jeremflow.com/projects/hindsight-hermes/</id>
    <summary>Why and how I added Hindsight, a long-term memory layer, on top of Hermes Agent — what it is, the stack, and how the memory flows.</summary>
  </entry>
  
  <entry>
    <title>Running Qwen 3.8 27B locally on an RTX 4090</title>
    <link href="https://jeremflow.com/notes/qwen3-8-27b-rtx4090/" />
    <updated>2026-06-02T20:45:00.000Z</updated>
    <id>https://jeremflow.com/notes/qwen3-8-27b-rtx4090/</id>
    <summary>Setup and observed performance of Qwen 3.8 27B (Q4_K_M) on a single RTX 4090 (24 GB), and why keeping it local matters.</summary>
  </entry>
  
  <entry>
    <title>The most efficient AI models right now (2026)</title>
    <link href="https://jeremflow.com/notes/efficient-ai-models/" />
    <updated>2026-05-25T18:30:00.000Z</updated>
    <id>https://jeremflow.com/notes/efficient-ai-models/</id>
    <summary>Which AI models give the best cost-to-quality ratio in 2026, the privacy caveat, and when not to use a cloud API.</summary>
  </entry>
  
  <entry>
    <title>Finding application notes fast with a targeted RAG</title>
    <link href="https://jeremflow.com/projects/rag-appnotes-strategy/" />
    <updated>2026-05-18T09:15:00.000Z</updated>
    <id>https://jeremflow.com/projects/rag-appnotes-strategy/</id>
    <summary>How I designed and deployed a small, private RAG on top of my application notes — the installed stack, the ingestion pipeline (Markdown, local VLM, indexing) and how I query it.</summary>
  </entry>
  
</feed>