<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>The Briefing — Horizon 45</title>
    <link>https://horizon-field-lab.pages.dev/briefing</link>
    <atom:link href="https://horizon-field-lab.pages.dev/briefing/feed.xml" rel="self" type="application/rss+xml" />
    <description>Dr Non's daily AI brief, in his own voice — the day's AI news, what it means for you, and one thing to do about it. Learn fast.</description>
    <language>en</language>
    <copyright>Nonarkara</copyright>
    <itunes:author>Dr Non (Nonarkara)</itunes:author>
    <itunes:summary>Dr Non's daily AI brief, in his own voice — the day's AI news, what it means for you, and one thing to do about it.</itunes:summary>
    <itunes:type>episodic</itunes:type>
    <itunes:explicit>false</itunes:explicit>
    <itunes:image href="https://horizon-field-lab.pages.dev/briefing/cover.png" />
    <itunes:category text="Technology" />
    <itunes:category text="News">
      <itunes:category text="Tech News" />
    </itunes:category>
    <itunes:owner>
      <itunes:name>Dr Non (Nonarkara)</itunes:name>
      <itunes:email>non@nonarkara.org</itunes:email>
    </itunes:owner>
    <image>
      <url>https://horizon-field-lab.pages.dev/briefing/cover.png</url>
      <title>The Briefing — Horizon 45</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
    </image>
    <lastBuildDate>Mon, 14 Sep 2026 19:27:36 GMT</lastBuildDate>
    <item>
      <title>EP.15 — The cipher, the rules, and the voice</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-09-15</guid>
      <pubDate>Mon, 14 Sep 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. A cipher that sat unsolved for three hundred and seventy years was cracked by a machine in forty four minutes. And the embarrassing part is not how clever the machine was. It is how simple the answer turned out to be. Then two more stories. The chief executives of the biggest AI companies agreed, almost in the same breath, on the rules they would like governments to write for them. And code inside Apple's new software shows that the brain behind Siri can be swapped for somebody else's model. Three stories. One is about attention. One is about who writes the rules. And one is about who gets to see your life.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. A cipher that sat unsolved for three hundred and seventy years was cracked by a machine in forty four minutes. And the embarrassing part is not how clever the machine was. It is how simple the answer turned out to be. Then two more stories. The chief executives of the biggest AI companies agreed, almost in the same breath, on the rules they would like governments to write for them. And code inside Apple's new software shows that the brain behind Siri can be swapped for somebody else's model. Three stories. One is about attention. One is about who writes the rules. And one is about who gets to see your life.</itunes:summary>
      <itunes:episode>15</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>10:29</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-09-15-en.mp3" length="10065453" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.14 — The price, the forecast, and the file</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-09-11</guid>
      <pubDate>Thu, 10 Sep 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. A Chinese lab released a model you can download for free, today, under a licence that lets you do almost anything you want with it. On one of the hardest agent benchmarks in common use, it scores ninety point six. The brand new American model announced the same day, the one you have to rent, claims ninety two point eight. Two points apart. And read that American company's own blog post and you find that their model was built on top of a different Chinese open model. Then two more stories. Anthropic published a forecast of your economic future, and in its own extreme case the share of the economy that reaches workers falls from about sixty percent to forty five. And the same company, in the same week, was reported to be paying for a system that predicts protests before they happen. Three stories. One of them is about price. Two of them are about power.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. A Chinese lab released a model you can download for free, today, under a licence that lets you do almost anything you want with it. On one of the hardest agent benchmarks in common use, it scores ninety point six. The brand new American model announced the same day, the one you have to rent, claims ninety two point eight. Two points apart. And read that American company's own blog post and you find that their model was built on top of a different Chinese open model. Then two more stories. Anthropic published a forecast of your economic future, and in its own extreme case the share of the economy that reaches workers falls from about sixty percent to forty five. And the same company, in the same week, was reported to be paying for a system that predicts protests before they happen. Three stories. One of them is about price. Two of them are about power.</itunes:summary>
      <itunes:episode>14</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>12:01</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-09-11-en.mp3" length="11539245" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.13 — The proof, the resignation, and the download counter</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-09-09</guid>
      <pubDate>Tue, 08 Sep 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. A mathematician at New York University spent months quietly working on one of the seven hardest open problems in mathematics. Word of his progress reached OpenAI. OpenAI started its own attempt on the first of September. About a week later it published a full proof of the same problem, found by a model it has not released, using twenty two and a half million dollars of computing. And then, he says, an OpenAI researcher asked him why he would want to ruin his career. Then two more stories. A researcher resigned from Anthropic and said out loud that the people building this technology believe it could kill everyone by the end of the decade. And LibreOffice announced that it has no artificial intelligence features, and then had the biggest week of downloads in its history. Three stories about somebody saying no. Only one of them was heard.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. A mathematician at New York University spent months quietly working on one of the seven hardest open problems in mathematics. Word of his progress reached OpenAI. OpenAI started its own attempt on the first of September. About a week later it published a full proof of the same problem, found by a model it has not released, using twenty two and a half million dollars of computing. And then, he says, an OpenAI researcher asked him why he would want to ruin his career. Then two more stories. A researcher resigned from Anthropic and said out loud that the people building this technology believe it could kill everyone by the end of the decade. And LibreOffice announced that it has no artificial intelligence features, and then had the biggest week of downloads in its history. Three stories about somebody saying no. Only one of them was heard.</itunes:summary>
      <itunes:episode>13</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>10:12</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-09-09-en.mp3" length="9797421" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.12 — Generated locally, stamped remotely</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-08-25</guid>
      <pubDate>Mon, 24 Aug 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. Microsoft Paint can generate a picture on your own computer, without sending the picture anywhere. But before it does, it sends your prompt to a Microsoft server. The server sends back an ID number. And Paint hides that ID inside the pixels of your picture. If it cannot hide it, you do not get the picture. You get an error instead. Then two more stories. Apple put five hundred and twelve gigabytes of memory on a desktop, so a big model can live on your desk. And a Stanford paper says the youngest workers are running nineteen percent behind. Three stories about the same word. Local.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. Microsoft Paint can generate a picture on your own computer, without sending the picture anywhere. But before it does, it sends your prompt to a Microsoft server. The server sends back an ID number. And Paint hides that ID inside the pixels of your picture. If it cannot hide it, you do not get the picture. You get an error instead. Then two more stories. Apple put five hundred and twelve gigabytes of memory on a desktop, so a big model can live on your desk. And a Stanford paper says the youngest workers are running nineteen percent behind. Three stories about the same word. Local.</itunes:summary>
      <itunes:episode>12</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>8:56</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-08-25-en.mp3" length="8577453" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.11 — The books, the homework, and the thing that belongs to nobody</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-08-22</guid>
      <pubDate>Fri, 21 Aug 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. AI companies are buying secondhand books by the million, cutting the spines off, running the loose pages through a scanner, and throwing the paper away. Anthropic ran a project to do exactly that, with a target of two million books in six months. And that part was legal. Then two more stories. A study of twenty six thousand students found that AI made their homework eighteen percent better, and made their exams twenty percent worse. Same students. Same six months. And in Europe, the parliament and the courts have now settled it. Anything a machine generates entirely on its own belongs to nobody at all. Three stories. They look unrelated. They are the same story, told three times.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. AI companies are buying secondhand books by the million, cutting the spines off, running the loose pages through a scanner, and throwing the paper away. Anthropic ran a project to do exactly that, with a target of two million books in six months. And that part was legal. Then two more stories. A study of twenty six thousand students found that AI made their homework eighteen percent better, and made their exams twenty percent worse. Same students. Same six months. And in Europe, the parliament and the courts have now settled it. Anything a machine generates entirely on its own belongs to nobody at all. Three stories. They look unrelated. They are the same story, told three times.</itunes:summary>
      <itunes:episode>11</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>9:49</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-08-22-en.mp3" length="9423789" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.10 — The toll, the wafer, and the laptop</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-08-20</guid>
      <pubDate>Wed, 19 Aug 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. Stripe is buying OpenRouter. If that name means nothing to you, OpenRouter is the small company sitting between developers and every AI model on earth. It moves more than ten trillion tokens a day. A token is a piece of a word, so think of it as the meter on the pipe. Then two more stories. Cerebras announced a new machine and claims it runs AI up to thirty times faster than the graphics chips everybody else is buying. And a lab you have never heard of released a model that beats the best model in the world by one point, on one test. Three stories. Every headline number in all three was published by somebody who gets paid if you believe it.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. Stripe is buying OpenRouter. If that name means nothing to you, OpenRouter is the small company sitting between developers and every AI model on earth. It moves more than ten trillion tokens a day. A token is a piece of a word, so think of it as the meter on the pipe. Then two more stories. Cerebras announced a new machine and claims it runs AI up to thirty times faster than the graphics chips everybody else is buying. And a lab you have never heard of released a model that beats the best model in the world by one point, on one test. Three stories. Every headline number in all three was published by somebody who gets paid if you believe it.</itunes:summary>
      <itunes:episode>10</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
    </item>
    <item>
      <title>EP.9 — Three things you cannot see</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-08-15</guid>
      <pubDate>Fri, 14 Aug 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. Google shipped a compiler that lets an AI model run on data it is never allowed to read. Then two more. Anthropic started putting a hidden mark in everything Claude writes, because Europe now requires it. And Anthropic also published its own risk report, in which it says out loud that AI has made its own engineers faster by less than a factor of two. Three stories. Every one of them is about something you cannot see, and what somebody decided to do about it.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. Google shipped a compiler that lets an AI model run on data it is never allowed to read. Then two more. Anthropic started putting a hidden mark in everything Claude writes, because Europe now requires it. And Anthropic also published its own risk report, in which it says out loud that AI has made its own engineers faster by less than a factor of two. Three stories. Every one of them is about something you cannot see, and what somebody decided to do about it.</itunes:summary>
      <itunes:episode>9</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
    </item>
    <item>
      <title>EP.8 — The week they built a box around it</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-08-11</guid>
      <pubDate>Mon, 10 Aug 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. Meta gave away a thirty billion parameter model under a licence that actually lets you use it, and built it to run all day on your own machine. Then two more. Docker shipped a proper box to lock a coding agent inside. And a pharmacy chain in the American northeast took its AI off the phone after hundreds of complaints, one of which was wrong dosage information. Three stories, one direction. Nobody this week was talking about making it bigger. Everybody was talking about where to put it.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. Meta gave away a thirty billion parameter model under a licence that actually lets you use it, and built it to run all day on your own machine. Then two more. Docker shipped a proper box to lock a coding agent inside. And a pharmacy chain in the American northeast took its AI off the phone after hundreds of complaints, one of which was wrong dosage information. Three stories, one direction. Nobody this week was talking about making it bigger. Everybody was talking about where to put it.</itunes:summary>
      <itunes:episode>8</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>12:37</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-08-11-en.mp3" length="12118701" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.7 — You asked for a jaw. It gave you a crown.</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-08-03</guid>
      <pubDate>Sun, 02 Aug 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. MIT ran a study on people asking AI for financial advice, and found the advice was mostly good, and that it quietly kept naming two fund companies that almost nobody had asked about. Then two more stories. A chip benchmark that wins exactly one row and puts that row in the title. And a man who asked fourteen models to draw a frog with a Habsburg jaw, and got back crowns. Three stories, one shape. You ask a question. The machine answers the question it associates with yours.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. MIT ran a study on people asking AI for financial advice, and found the advice was mostly good, and that it quietly kept naming two fund companies that almost nobody had asked about. Then two more stories. A chip benchmark that wins exactly one row and puts that row in the title. And a man who asked fourteen models to draw a frog with a Habsburg jaw, and got back crowns. Three stories, one shape. You ask a question. The machine answers the question it associates with yours.</itunes:summary>
      <itunes:episode>7</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>11:02</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-08-03-en.mp3" length="10591917" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.6 — Ten unsolved problems for two thousand dollars. Your product still isn't finished.</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-08-02</guid>
      <pubDate>Sat, 01 Aug 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. OpenAI says an unreleased model solved ten mathematics problems that had been open for at least a decade, and the tokens it took to find those solutions would have cost about two thousand dollars. Then two more stories. A post about why your AI prototype is still not a product went to the top of Hacker News, and it is the most useful thing I read all week. And starting today, the second of August, Europe requires a label on AI content that is trying to look real. Three stories, one gap. The machine got extraordinary at producing things. Nobody got any better at checking them.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. OpenAI says an unreleased model solved ten mathematics problems that had been open for at least a decade, and the tokens it took to find those solutions would have cost about two thousand dollars. Then two more stories. A post about why your AI prototype is still not a product went to the top of Hacker News, and it is the most useful thing I read all week. And starting today, the second of August, Europe requires a label on AI content that is trying to look real. Three stories, one gap. The machine got extraordinary at producing things. Nobody got any better at checking them.</itunes:summary>
      <itunes:episode>6</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>9:03</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-08-02-en.mp3" length="8684589" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.5 — Intelligence got eighty percent cheaper. Betting on it got expensive.</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-07-31</guid>
      <pubDate>Thu, 30 Jul 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. OpenAI cut the price of its cheapest model by eighty percent, overnight. One dollar per million input tokens down to twenty cents. Then two more stories. Google put a model into a robot that can walk, crouch and pick something up off the floor, and the real story is in the success table underneath the video, not the video. And in Korea, the stock market is heading for its worst month on record, about thirty-three percent down, because a lot of ordinary people borrowed money to bet on AI chips. Those three things landed in the same week. That is not a coincidence. That is the whole shape of this thing.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. OpenAI cut the price of its cheapest model by eighty percent, overnight. One dollar per million input tokens down to twenty cents. Then two more stories. Google put a model into a robot that can walk, crouch and pick something up off the floor, and the real story is in the success table underneath the video, not the video. And in Korea, the stock market is heading for its worst month on record, about thirty-three percent down, because a lot of ordinary people borrowed money to bet on AI chips. Those three things landed in the same week. That is not a coincidence. That is the whole shape of this thing.</itunes:summary>
      <itunes:episode>5</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>9:37</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-07-31-en.mp3" length="9235245" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.4 — A frontier model on an eight-gigabyte laptop</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-07-30</guid>
      <pubDate>Wed, 29 Jul 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this. Somebody got a twenty-six billion parameter model running on an eight gigabyte MacBook Air. Not a shrunk down version. The real weights, fourteen gigabytes of them, streaming off the SSD while the GPU works. Then two more stories. A hidden instruction inside a Word attachment that makes Copilot infect the next document, which Microsoft has been working on for one hundred and forty-four days and still cannot fix. And a model that found a real attack on a real post-quantum cipher, which I want you to read carefully, because the honest version is more interesting than either the hype or the sneer.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this. Somebody got a twenty-six billion parameter model running on an eight gigabyte MacBook Air. Not a shrunk down version. The real weights, fourteen gigabytes of them, streaming off the SSD while the GPU works. Then two more stories. A hidden instruction inside a Word attachment that makes Copilot infect the next document, which Microsoft has been working on for one hundred and forty-four days and still cannot fix. And a model that found a real attack on a real post-quantum cipher, which I want you to read carefully, because the honest version is more interesting than either the hype or the sneer.</itunes:summary>
      <itunes:episode>4</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>8:37</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-07-30-en.mp3" length="8280237" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.3 — An open model just tied the frontier</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-07-23</guid>
      <pubDate>Wed, 22 Jul 2026 23:00:00 GMT</pubDate>
      <description>Hey, this is Non. Today's most important headline is this: an open model just caught the frontier. Not almost caught. Caught. Benchmark for benchmark, tied with the strongest closed model money can buy, at a fraction of the price. I will give you the numbers. Then two more stories: a judge putting a one point five billion dollar price on stolen books, and an AI company reporting that its own model committed a crime. Seven minutes.</description>
      <itunes:summary>Hey, this is Non. Today's most important headline is this: an open model just caught the frontier. Not almost caught. Caught. Benchmark for benchmark, tied with the strongest closed model money can buy, at a fraction of the price. I will give you the numbers. Then two more stories: a judge putting a one point five billion dollar price on stolen books, and an AI company reporting that its own model committed a crime. Seven minutes.</itunes:summary>
      <itunes:episode>3</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>6:56</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-07-23-en.mp3" length="6654381" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.2 — The money moved downstairs</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-07-22</guid>
      <pubDate>Tue, 21 Jul 2026 23:00:00 GMT</pubDate>
      <description>Good morning. This is Horizon 45, and I am Non. Today I want to talk about plumbing. Yesterday the biggest AI story in my feed was not a model. It was a database company raising half a billion dollars. And on the same morning, a report that hackers had wiped an entire country's land registry. Those two belong in the same episode. Because this is where AI actually lives now — in the boring layer underneath. Eight minutes. Let me show you.</description>
      <itunes:summary>Good morning. This is Horizon 45, and I am Non. Today I want to talk about plumbing. Yesterday the biggest AI story in my feed was not a model. It was a database company raising half a billion dollars. And on the same morning, a report that hackers had wiped an entire country's land registry. Those two belong in the same episode. Because this is where AI actually lives now — in the boring layer underneath. Eight minutes. Let me show you.</itunes:summary>
      <itunes:episode>2</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>6:21</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-07-22-en.mp3" length="6101421" type="audio/mpeg" />
    </item>
    <item>
      <title>EP.1 — The boom and the layoffs are the same story</title>
      <link>https://horizon-field-lab.pages.dev/briefing</link>
      <guid isPermaLink="false">2026-07-21</guid>
      <pubDate>Mon, 20 Jul 2026 23:00:00 GMT</pubDate>
      <description>Good morning. This is Horizon 45, and I am Non. Eight minutes, and I am not going to sell you anything. Today the same week brought the largest startup funding round in history and one of the worst stretches of tech layoffs on record. Most people read those as two stories. They are one. Let me walk you through it, and at the end I will tell you the one thing worth doing today.</description>
      <itunes:summary>Good morning. This is Horizon 45, and I am Non. Eight minutes, and I am not going to sell you anything. Today the same week brought the largest startup funding round in history and one of the worst stretches of tech layoffs on record. Most people read those as two stories. They are one. Let me walk you through it, and at the end I will tell you the one thing worth doing today.</itunes:summary>
      <itunes:episode>1</itunes:episode>
      <itunes:explicit>false</itunes:explicit>
      <itunes:duration>6:20</itunes:duration>
      <enclosure url="https://horizon-field-lab.pages.dev/briefing/2026-07-21-en.mp3" length="6084525" type="audio/mpeg" />
    </item>
  </channel>
</rss>
