<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel>
    <title>muun notes</title>
    <link>https://muun.agency/notes/</link>
    <description>Notes and articles on automation, AI and the tools we build.</description>
    <language>en</language>
    <item>
      <title>Tooltip | Choose the right model for the job</title>
      <link>https://muun.agency/notes/choose-right-model-for-job/</link>
      <guid isPermaLink="false">2026-10-08-choose-right-model-for-job</guid>
      <pubDate>Thu, 08 Oct 2026 17:00:00 GMT</pubDate>
      <description><![CDATA[<p>I won&#39;t go into details about which model to use when, but every model has its strengths, and choosing the right model for the job is part of the job of using AI well.</p>
<p>If you are looking for a quick and easy way to test different models for your app or use case, try <a href="https://openrouter.ai/">OpenRouter</a>. You only need one account and one API key, and you get access to numerous models (by OpenAI, Anthropic, Google, Meta and others). Switching between models is as easy as changing the model name. Testing isn&#39;t the only reason to use OpenRouter. Models are constantly evolving, new ones are popping up, and sometimes existing models experience downtime, and you need to make a quick switch to continue production.</p>
]]></description>
    </item>
    <item>
      <title>HOW I AI | Why I'm reluctant to give Claude and ChatGPT access to my emails</title>
      <link>https://muun.agency/notes/ai-access-to-my-emails/</link>
      <guid isPermaLink="false">2026-10-08-ai-access-to-my-emails</guid>
      <pubDate>Thu, 08 Oct 2026 14:00:00 GMT</pubDate>
      <description><![CDATA[<p>One of the gateways to incorporating AI into daily life is giving it access to emails. I know people use it to stay organized, create task lists, send emails on their behalf, etc. I am tempted, but I am reluctant to give agents (apart from the one my email provider already uses) access to my emails because:</p>
<ol>
<li>emails can contain private information,</li>
<li>the other party has not given me permission to share their info with companies like OpenAI and Anthropic, and</li>
<li>I want to limit access and exposure to my personal information (whenever possible).</li>
</ol>
<p>I&#39;m not perfect, but I make a concerted effort to minimize the amount of personal information I share with models. When conducting research or analysis, I scrub, depersonalize, and redact as much as I can, and stick to publicly available information and data if identity is necessary for context.</p>
<p>Unfortunately, not everyone shares the same level of caution, and I imagine my personal details sent over email are floating around on the interwebs and being used to train models without my explicit consent.</p>
]]></description>
    </item>
    <item>
      <title>PSA | Note to companies, influencers, thought leaders, webinar hosts</title>
      <link>https://muun.agency/notes/unsolicited-calendar-invites/</link>
      <guid isPermaLink="false">2026-10-05-unsolicited-calendar-invites</guid>
      <pubDate>Mon, 05 Oct 2026 20:00:00 GMT</pubDate>
      <description><![CDATA[<p>I&#39;m seeing more and more random notices pop up in my digital calendar. I think what companies and influencers are doing is sending calendar invites to their list, which causes them to pop up in my calendar to be accepted or declined.</p>
<p>The first time it happened, I was confused. I sign up for so many online events. Sometimes I don&#39;t remember what I signed up for, so I attended. Turned out to be a sales pitch and a complete waste of my time. Please don&#39;t do this. It&#39;s confusing at best, and feels gross and violating at worst.</p>
]]></description>
    </item>
    <item>
      <title>CLAUDE WHISPERER | How I get good results from Claude</title>
      <link>https://muun.agency/notes/good-results-from-claude/</link>
      <guid isPermaLink="false">2026-10-05-good-results-from-claude</guid>
      <pubDate>Mon, 05 Oct 2026 14:00:00 GMT</pubDate>
      <description><![CDATA[<ol>
<li><strong>Have a clear plan or goal and stay laser-focused on it (does not apply to brainstorms)</strong> - This helps prevent getting distracted by all the shiny objects (i.e., ideas and all the words Claude and other LLMs are happy to give you). I also share my goal with Claude at the start—and remind it along the way.</li>
<li><strong>Conduct pilots</strong> - When I&#39;m working on a large, complex project, I test smaller ideas before rolling out across the project. This does a few things: clarifies my thoughts, refines my approach, and limits the blast radius of mistakes. By preventing errors or bad ideas from being applied at scale, I reduce the number of rollbacks and save tokens.</li>
<li><strong>Perform sweeps</strong> - For multi-layer or multi-faceted projects, rather than perfect every idea before moving on to the next, I get the basic ideas down, then refine using layered sweeps. I give Claude 1–3 tasks to perform across the entire project or body of work. Then, rinse and repeat (as needed). This is how I get consistent and cohesive results. Also, this generally reduces the number of LLM calls.</li>
</ol>
]]></description>
    </item>
</channel></rss>
