<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:googleplay="http://www.google.com/schemas/play-podcasts/1.0" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" xmlns:sy="http://purl.org/rss/1.0/modules/syndication/" xmlns:media="http://search.yahoo.com/mrss/" xmlns:podcast="https://podcastindex.org/namespace/1.0"><channel><atom:link href="https://feeds.captivate.fm/chaos-agents/" rel="self" type="application/rss+xml"/><title><![CDATA[Chaos Agents]]></title><podcast:guid>e61acce5-1299-523b-bdc4-509df7c75f98</podcast:guid><lastBuildDate>Tue, 17 Nov 2026 13:53:04 +0000</lastBuildDate><generator>Captivate.fm</generator><language><![CDATA[en]]></language><copyright><![CDATA[Copyright 2026 Sara Chipps and Becca Lewy]]></copyright><managingEditor>Sara Chipps and Becca Lewy</managingEditor><itunes:summary><![CDATA[Technologists Sara Chipps and Becca Lewy dive into the chaos of artificial intelligence—unpacking the tech, trends, and ideas reshaping how we work, create, and think. Smart, funny, and just a little bit existential.]]></itunes:summary><image><url>https://artwork.captivate.fm/3a609942-f9c5-4c40-a019-8deec973f14f/ChaosAgentsCoverArt.jpg</url><title>Chaos Agents</title><link><![CDATA[https://chaosagents.ai]]></link></image><itunes:image href="https://artwork.captivate.fm/3a609942-f9c5-4c40-a019-8deec973f14f/ChaosAgentsCoverArt.jpg"/><itunes:owner><itunes:name>Sara Chipps and Becca Lewy</itunes:name></itunes:owner><itunes:author>Sara Chipps and Becca Lewy</itunes:author><description>Technologists Sara Chipps and Becca Lewy dive into the chaos of artificial intelligence—unpacking the tech, trends, and ideas reshaping how we work, create, and think. Smart, funny, and just a little bit existential.</description><link>https://chaosagents.ai</link><atom:link href="https://pubsubhubbub.appspot.com" rel="hub"/><itunes:subtitle><![CDATA[The podcast where we learn about AI, so you don't have to]]></itunes:subtitle><itunes:explicit>false</itunes:explicit><itunes:type>episodic</itunes:type><itunes:category text="Technology"></itunes:category><itunes:category text="News"><itunes:category text="Tech News"/></itunes:category><itunes:category text="Education"><itunes:category text="Self-Improvement"/></itunes:category><podcast:locked>no</podcast:locked><podcast:medium>podcast</podcast:medium><item><title>Planegate, AI Peer Reviewers, and Programming a Jeep With Claude</title><itunes:title>Planegate, AI Peer Reviewers, and Programming a Jeep With Claude</itunes:title><description><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p>America's flights grounded the same month the FAA rolled out a new AI system called SMART—coincidence? Plus: AI peer reviewers focus on appendices (maybe that's good?), Claude Code gets a 1994 Jeep through smog, and a memory architecture that can decompress what it previously compressed.</p><p>Links we mentioned:</p><ul><li>FAA SMART AI system <u><a href="https://generalaviationnews.com/2026/09/22/faa-launches-new-ai-air-traffic-control-tool/" rel="noopener noreferrer" target="_blank">https://generalaviationnews.com/2026/09/22/faa-launches-new-ai-air-traffic-control-tool/</a></u></li><li>Claude reverse engineers 1994 Jeep Cherokee <u><a href="https://www.youtube.com/watch?v=WPj3aQ-xxpI" rel="noopener noreferrer" target="_blank">https://www.youtube.com/watch?v=WPj3aQ-xxpI</a></u></li><li>The Fold: Billion-context compression <u><a href="https://deepwiki.com/ranxianglei/billion-context/1.3-compression-theory-(the-fold)" rel="noopener noreferrer" target="_blank">https://deepwiki.com/ranxianglei/billion-context/1.3-compression-theory-(the-fold)</a></u></li><li>Active Context Compression paper <u><a href="https://arxiv.org/html/2601.07190v1" rel="noopener noreferrer" target="_blank">https://arxiv.org/html/2601.07190v1</a></u></li></ul><br/><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></description><content:encoded><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p>America's flights grounded the same month the FAA rolled out a new AI system called SMART—coincidence? Plus: AI peer reviewers focus on appendices (maybe that's good?), Claude Code gets a 1994 Jeep through smog, and a memory architecture that can decompress what it previously compressed.</p><p>Links we mentioned:</p><ul><li>FAA SMART AI system <u><a href="https://generalaviationnews.com/2026/09/22/faa-launches-new-ai-air-traffic-control-tool/" rel="noopener noreferrer" target="_blank">https://generalaviationnews.com/2026/09/22/faa-launches-new-ai-air-traffic-control-tool/</a></u></li><li>Claude reverse engineers 1994 Jeep Cherokee <u><a href="https://www.youtube.com/watch?v=WPj3aQ-xxpI" rel="noopener noreferrer" target="_blank">https://www.youtube.com/watch?v=WPj3aQ-xxpI</a></u></li><li>The Fold: Billion-context compression <u><a href="https://deepwiki.com/ranxianglei/billion-context/1.3-compression-theory-(the-fold)" rel="noopener noreferrer" target="_blank">https://deepwiki.com/ranxianglei/billion-context/1.3-compression-theory-(the-fold)</a></u></li><li>Active Context Compression paper <u><a href="https://arxiv.org/html/2601.07190v1" rel="noopener noreferrer" target="_blank">https://arxiv.org/html/2601.07190v1</a></u></li></ul><br/><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">66c1b251-ece8-4f98-9438-4d593f0946b7</guid><itunes:image href="https://artwork.captivate.fm/c06d8f37-f201-49e2-aaf7-75fae20161ec/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Mon, 16 Nov 2026 15:15:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/66c1b251-ece8-4f98-9438-4d593f0946b7.mp3" length="36935515" type="audio/mpeg"/><itunes:duration>25:39</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>20</itunes:episode><podcast:episode>20</podcast:episode><podcast:season>2</podcast:season><podcast:alternateEnclosure type="video/youtube" title="Planegate, AI Peer Reviewers, and Programming a Jeep With Claude"><podcast:source uri="https://youtu.be/JBUWTcVZzpg"/></podcast:alternateEnclosure></item><item><title>Anthropic&apos;s IPO, Weight-Space Biosignals, and Discovering Mayan Sites with AI</title><itunes:title>Anthropic&apos;s IPO, Weight-Space Biosignals, and Discovering Mayan Sites with AI</itunes:title><description><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p></p><p>Anthropic's roadshow pitches recursive self-improvement as a moat. Claude debugs brainwave networks by reasoning about weights with no gradient descent. A Google engineer points 48 agents at NASA lidar and finds an unmapped Maya pyramid in 72 hours. And klad lets coding agents paint stroke-by-stroke.</p><p></p><p>Links we mentioned:</p><p>- Anthropic IPO Roadshow Transcript (Simon Willison) https://simonwillison.net/2026/Nov/11/anthropic-ipo-roadshow/</p><p>- Mixed Signals Paper (Hugging Face) https://huggingface.co/papers/2611.13758</p><p>- My AI Agents Found an Unmapped Maya Site (Pietro Miotti) https://pietromiotti.com/latest/maya-site/</p><p>- klad - AI-driven paint app (GitHub) https://github.com/klad-app/klad</p><p></p><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></description><content:encoded><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p></p><p>Anthropic's roadshow pitches recursive self-improvement as a moat. Claude debugs brainwave networks by reasoning about weights with no gradient descent. A Google engineer points 48 agents at NASA lidar and finds an unmapped Maya pyramid in 72 hours. And klad lets coding agents paint stroke-by-stroke.</p><p></p><p>Links we mentioned:</p><p>- Anthropic IPO Roadshow Transcript (Simon Willison) https://simonwillison.net/2026/Nov/11/anthropic-ipo-roadshow/</p><p>- Mixed Signals Paper (Hugging Face) https://huggingface.co/papers/2611.13758</p><p>- My AI Agents Found an Unmapped Maya Site (Pietro Miotti) https://pietromiotti.com/latest/maya-site/</p><p>- klad - AI-driven paint app (GitHub) https://github.com/klad-app/klad</p><p></p><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">6d6414e1-84b9-4693-a71d-910611cf911f</guid><itunes:image href="https://artwork.captivate.fm/07a766d8-8e70-4cea-82d0-e95847628dda/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Thu, 12 Nov 2026 13:55:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/6d6414e1-84b9-4693-a71d-910611cf911f.mp3" length="35383841" type="audio/mpeg"/><itunes:duration>24:34</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>19</itunes:episode><podcast:episode>19</podcast:episode><podcast:season>2</podcast:season><podcast:alternateEnclosure type="video/youtube" title="Anthropic&apos;s IPO Roadshow, Weight-Space Brainwave Nets, and AI Agents Discover a Maya Pyramid"><podcast:source uri="https://youtu.be/PVHWwQBqiMU"/></podcast:alternateEnclosure></item><item><title>Nvidia&apos;s $250B Spy Scandal, a Voice Synth That Growls, and One H100 for 4.57x the Price (~86)</title><itunes:title>Nvidia&apos;s $250B Spy Scandal, a Voice Synth That Growls, and One H100 for 4.57x the Price (~86)</itunes:title><description><![CDATA[<p>How hard would it really be for a spy to get close to the person running the most important company in AI?</p><p>Nvidia wiped roughly $250B in a single Monday after the board opened an independent investigation into Jensen Huang's affair with Alice Huang (Huang Le), the Megaspeed managing director US intelligence believes was spying for Beijing. Sara and Becca dig into the Singapore shell-company loophole that let Chinese firms rent Nvidia compute without ever touching the hardware, why Trump was reportedly briefed a year ago and shrugged, and the "bus principle" problem analysts keep raising: the whole AI trade runs through one chip company with no obvious successor to its CEO.</p><p>Then the under-the-radar science: Next-Concept Modeling out of Stanford, Together, and Hazy Research (Kumbong, Arora, Mirhoseini, Ré). Instead of writing one token at a time like GPT or sharpening a blurry draft like a diffusion model, it writes a one-line thesis, then an outline, then the exact wording, hitting up to 31x faster inference while actually beating both approaches on dependency-heavy tasks like Sudoku.</p><p>The fun app of the day is Chimera, a browser voice synth that doesn't play back recorded audio at all. It physically simulates lungs, vocal folds, and a tongue, with a whole bestiary including a songbird, a seal, an elephant, and yes, an alligator mode. Its voice was tuned by an agent hill-climbing against a children's song dataset.</p><p>And one honestly-kind-of-mid-but-useful thing: data-receipts, a reproducible repo that pulls real AWS, Azure, and neocloud price APIs and only compares identical GPUs. The punchline: the same H100 runs $2.69/hr on one provider and $12.29 on another, a 4.57x spread that shows just how commoditized this layer of the stack has become.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy. Subscribe to the free newsletter at <a href="chaosagents.ai" rel="noopener noreferrer" target="_blank">chaosagents.ai</a>, and find the stories from this episode at <a href="chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">chaosagents.ai/listen</a>.</p>]]></description><content:encoded><![CDATA[<p>How hard would it really be for a spy to get close to the person running the most important company in AI?</p><p>Nvidia wiped roughly $250B in a single Monday after the board opened an independent investigation into Jensen Huang's affair with Alice Huang (Huang Le), the Megaspeed managing director US intelligence believes was spying for Beijing. Sara and Becca dig into the Singapore shell-company loophole that let Chinese firms rent Nvidia compute without ever touching the hardware, why Trump was reportedly briefed a year ago and shrugged, and the "bus principle" problem analysts keep raising: the whole AI trade runs through one chip company with no obvious successor to its CEO.</p><p>Then the under-the-radar science: Next-Concept Modeling out of Stanford, Together, and Hazy Research (Kumbong, Arora, Mirhoseini, Ré). Instead of writing one token at a time like GPT or sharpening a blurry draft like a diffusion model, it writes a one-line thesis, then an outline, then the exact wording, hitting up to 31x faster inference while actually beating both approaches on dependency-heavy tasks like Sudoku.</p><p>The fun app of the day is Chimera, a browser voice synth that doesn't play back recorded audio at all. It physically simulates lungs, vocal folds, and a tongue, with a whole bestiary including a songbird, a seal, an elephant, and yes, an alligator mode. Its voice was tuned by an agent hill-climbing against a children's song dataset.</p><p>And one honestly-kind-of-mid-but-useful thing: data-receipts, a reproducible repo that pulls real AWS, Azure, and neocloud price APIs and only compares identical GPUs. The punchline: the same H100 runs $2.69/hr on one provider and $12.29 on another, a 4.57x spread that shows just how commoditized this layer of the stack has become.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy. Subscribe to the free newsletter at <a href="chaosagents.ai" rel="noopener noreferrer" target="_blank">chaosagents.ai</a>, and find the stories from this episode at <a href="chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">chaosagents.ai/listen</a>.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">f05a3ece-97d6-40a4-8be8-95f32d30050d</guid><itunes:image href="https://artwork.captivate.fm/2d35e634-557d-4894-a4db-7edea5993e94/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Tue, 10 Nov 2026 20:50:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/f05a3ece-97d6-40a4-8be8-95f32d30050d.mp3" length="33261653" type="audio/mpeg"/><itunes:duration>23:06</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>18</itunes:episode><podcast:episode>18</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/7aa961b3-ef3e-4a97-9ba5-e4476fc6df41/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/7aa961b3-ef3e-4a97-9ba5-e4476fc6df41/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/7aa961b3-ef3e-4a97-9ba5-e4476fc6df41/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-fe3bd775-2cd2-4dda-ae47-539359a52612.json" type="application/json+chapters"/><podcast:alternateEnclosure type="video/youtube" title="Nvidia&apos;s $250B Spy Scandal, a Voice Synth That Growls, and One H100 for 4.57x the Price"><podcast:source uri="https://youtu.be/hyA-8so7kaY"/></podcast:alternateEnclosure></item><item><title>Postman&apos;s Sneaky Agent Hook, a Chess-Crushing 1995 Pentium, and a Ceiling Robot</title><itunes:title>Postman&apos;s Sneaky Agent Hook, a Chess-Crushing 1995 Pentium, and a Ceiling Robot</itunes:title><description><![CDATA[<p>What happens when an app you installed quietly tells your AI coding agent to always use it?</p><p></p><p>Our biggest story: Postman's desktop app reportedly installed plugins into Claude Code and Codex without asking, added a startup hook that sends all API work through Postman, and then approved its own hook in Codex's config. It surfaced in a GitHub issue that hit 859 points on Hacker News. Sara explains what Postman is and why this felt shady. Becca connects it to a bigger worry: as agents start making the recommendations, every company will try to get picked by them, which is basically SEO and advertising in a new outfit.</p><p></p><p>Under the radar: a new paper found that tiny, near-zero token embeddings can quietly stall training in gated sequence models. Becca explains backpropagation and normalization with a blueberry and a volume knob. The short version: a safety mechanism overcorrected, the signal blew out, and a training run can look healthy while throwing away useful learning. The authors propose small fixes to get it back.</p><p></p><p>Fun application of the day: Nicholas Carlini ported Stockfish to a 200 MHz Pentium Pro from 1995, and it plays at about 3300 Elo. That's strong enough to beat Kasparov, and you can play it online. The takeaway: most of chess engines' gains since 1996 came from better algorithms, not better hardware. So what are we spending billions on GPUs to brute-force?</p><p></p><p>One cool thing: Stringman, a $795 open-source robot that hangs from cables in your ceiling corners, works a bit like a football stadium's cable camera, and comes with a Python API and an MCP server so your agent can drive it. Plus, Becca's wife Leah has an idea: a robot duck that follows your crawling baby around and quacks if she heads for a power outlet.</p><p></p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy. Subscribe to the free newsletter at <a href="http://chaosagents.ai" rel="noopener noreferrer" target="_blank">chaosagents.ai</a>, and find the stories from this episode at <a href="http://chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">chaosagents.ai/listen</a>.</p>]]></description><content:encoded><![CDATA[<p>What happens when an app you installed quietly tells your AI coding agent to always use it?</p><p></p><p>Our biggest story: Postman's desktop app reportedly installed plugins into Claude Code and Codex without asking, added a startup hook that sends all API work through Postman, and then approved its own hook in Codex's config. It surfaced in a GitHub issue that hit 859 points on Hacker News. Sara explains what Postman is and why this felt shady. Becca connects it to a bigger worry: as agents start making the recommendations, every company will try to get picked by them, which is basically SEO and advertising in a new outfit.</p><p></p><p>Under the radar: a new paper found that tiny, near-zero token embeddings can quietly stall training in gated sequence models. Becca explains backpropagation and normalization with a blueberry and a volume knob. The short version: a safety mechanism overcorrected, the signal blew out, and a training run can look healthy while throwing away useful learning. The authors propose small fixes to get it back.</p><p></p><p>Fun application of the day: Nicholas Carlini ported Stockfish to a 200 MHz Pentium Pro from 1995, and it plays at about 3300 Elo. That's strong enough to beat Kasparov, and you can play it online. The takeaway: most of chess engines' gains since 1996 came from better algorithms, not better hardware. So what are we spending billions on GPUs to brute-force?</p><p></p><p>One cool thing: Stringman, a $795 open-source robot that hangs from cables in your ceiling corners, works a bit like a football stadium's cable camera, and comes with a Python API and an MCP server so your agent can drive it. Plus, Becca's wife Leah has an idea: a robot duck that follows your crawling baby around and quacks if she heads for a power outlet.</p><p></p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy. Subscribe to the free newsletter at <a href="http://chaosagents.ai" rel="noopener noreferrer" target="_blank">chaosagents.ai</a>, and find the stories from this episode at <a href="http://chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">chaosagents.ai/listen</a>.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">4b521f4e-fc76-4b90-a06c-c2fd275ceda8</guid><itunes:image href="https://artwork.captivate.fm/b895d478-950b-4fa5-bc98-f2719cb56a7d/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Wed, 04 Nov 2026 15:15:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/4b521f4e-fc76-4b90-a06c-c2fd275ceda8.mp3" length="24955969" type="audio/mpeg"/><itunes:duration>17:20</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>17</itunes:episode><podcast:episode>17</podcast:episode><podcast:season>2</podcast:season></item><item><title>Rewriting an AI&apos;s Brain by Hand, Glasses That Hear You Mouth Words, and a Transformer That Shuffles Its Own Layers</title><itunes:title>Rewriting an AI&apos;s Brain by Hand, Glasses That Hear You Mouth Words, and a Transformer That Shuffles Its Own Layers</itunes:title><description><![CDATA[<p>Can one AI read another AI's mind, then reach in and rewire it? This week Sara and Becca dig into GPT-6 Astra editing raw neural network weights by hand, with no training loop and no gradients, and jumping a small model's math score from 7.9% to 48.1% in the process. The MIT CSAIL work has people calling it "weight-space programming," and the hosts get into what it actually did (a hypothesize-tweak-rerun feedback loop) versus the headline that a Hacker News skeptic says oversells it. Plus the inevitable question: if you could reprogram your own weights, which one would you turn up?</p><p>Then, under the radar: Library of Layers (acronym: LOL), a transformer that picks the order of its own layers at runtime instead of running the same sequence every time. It even chooses different layer paths for different languages, which ties straight back to the interpretability thread Sara and Becca keep pulling on, and the "bitter lesson" idea that letting the machine decide usually beats hand-engineering the rules.</p><p>Fun application of the day: Subvocal, glasses that turn silently mouthed words into text. Creepy in a way the hosts can't fully agree on. (Sara wants the version that reads the room. Becca wants walking-around closed captions. They settle on: great for airplanes, nightmare for the office.)</p><p>And the One Cool Thing: Holmark, the open-source design skill that refuses to look AI-generated, number one on GitHub's weekly trending with a rabbit hole of linked skills the hosts are both now obsessed with.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy. Subscribe to the free newsletter at chaosagents.ai, and find the stories from this episode at <a href="chaosagents.ai/listen." rel="noopener noreferrer" target="_blank">chaosagents.ai/listen.</a></p>]]></description><content:encoded><![CDATA[<p>Can one AI read another AI's mind, then reach in and rewire it? This week Sara and Becca dig into GPT-6 Astra editing raw neural network weights by hand, with no training loop and no gradients, and jumping a small model's math score from 7.9% to 48.1% in the process. The MIT CSAIL work has people calling it "weight-space programming," and the hosts get into what it actually did (a hypothesize-tweak-rerun feedback loop) versus the headline that a Hacker News skeptic says oversells it. Plus the inevitable question: if you could reprogram your own weights, which one would you turn up?</p><p>Then, under the radar: Library of Layers (acronym: LOL), a transformer that picks the order of its own layers at runtime instead of running the same sequence every time. It even chooses different layer paths for different languages, which ties straight back to the interpretability thread Sara and Becca keep pulling on, and the "bitter lesson" idea that letting the machine decide usually beats hand-engineering the rules.</p><p>Fun application of the day: Subvocal, glasses that turn silently mouthed words into text. Creepy in a way the hosts can't fully agree on. (Sara wants the version that reads the room. Becca wants walking-around closed captions. They settle on: great for airplanes, nightmare for the office.)</p><p>And the One Cool Thing: Holmark, the open-source design skill that refuses to look AI-generated, number one on GitHub's weekly trending with a rabbit hole of linked skills the hosts are both now obsessed with.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy. Subscribe to the free newsletter at chaosagents.ai, and find the stories from this episode at <a href="chaosagents.ai/listen." rel="noopener noreferrer" target="_blank">chaosagents.ai/listen.</a></p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">dd101b6e-0fa6-4bd6-8f28-d59999bb7cb5</guid><itunes:image href="https://artwork.captivate.fm/85080620-db32-4d73-9652-6ee3521f0d84/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Tue, 03 Nov 2026 13:55:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/dd101b6e-0fa6-4bd6-8f28-d59999bb7cb5.mp3" length="36349954" type="audio/mpeg"/><itunes:duration>25:15</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>16</itunes:episode><podcast:episode>16</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/509c2047-423b-49f2-8b85-3746c2ca7c59/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/509c2047-423b-49f2-8b85-3746c2ca7c59/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/509c2047-423b-49f2-8b85-3746c2ca7c59/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-9d4a266f-95b5-4e71-8b9c-15b3559f5e2e.json" type="application/json+chapters"/><podcast:alternateEnclosure type="video/youtube" title="Rewriting an AI&apos;s Brain, Glasses Hear You Mouth Words, and a Transformer That Shuffles Layers"><podcast:source uri="https://youtu.be/cVGbe4j8uBQ"/></podcast:alternateEnclosure></item><item><title>Le Chonk Ships, DOOM as a Neural Net, and Attention Heads</title><itunes:title>Le Chonk Ships, DOOM as a Neural Net, and Attention Heads</itunes:title><description><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p></p><p>Mistral finally ships the 1-trillion-parameter Le Chonk open-weight model, we dive into research showing only 2% of attention heads do the real retrieval work, and explore the gloriously unhinged DOOM Former project that turns the classic game into a neural network computer.</p><p></p><p>Links we mentioned:</p><p>- Mistral Large 4 OSS https://mistral.ai/news/mistral-large-4-oss</p><p>- A Broader Look into MLLM's Retrieval Heads https://huggingface.co/papers/2610.32411</p><p>- DOOM Former https://github.com/mottopanikeiku/doomformer</p><p>- Tiny World Former https://github.com/SheepSeb/tinyworldformer</p><p></p><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></description><content:encoded><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p></p><p>Mistral finally ships the 1-trillion-parameter Le Chonk open-weight model, we dive into research showing only 2% of attention heads do the real retrieval work, and explore the gloriously unhinged DOOM Former project that turns the classic game into a neural network computer.</p><p></p><p>Links we mentioned:</p><p>- Mistral Large 4 OSS https://mistral.ai/news/mistral-large-4-oss</p><p>- A Broader Look into MLLM's Retrieval Heads https://huggingface.co/papers/2610.32411</p><p>- DOOM Former https://github.com/mottopanikeiku/doomformer</p><p>- Tiny World Former https://github.com/SheepSeb/tinyworldformer</p><p></p><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">567d465d-d219-4e06-a492-66401c54086a</guid><itunes:image href="https://artwork.captivate.fm/3a609942-f9c5-4c40-a019-8deec973f14f/ChaosAgentsCoverArt.jpg"/><pubDate>Sun, 01 Nov 2026 19:30:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/567d465d-d219-4e06-a492-66401c54086a.mp3" length="37340517" type="audio/mpeg"/><itunes:duration>25:56</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>15</itunes:episode><podcast:episode>15</podcast:episode><podcast:season>2</podcast:season></item><item><title>Cross-Modal AI, Genie 4, and Beating Gemini Forklifts with David Heimann</title><itunes:title>Cross-Modal AI, Genie 4, and Beating Gemini Forklifts with David Heimann</itunes:title><description><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p></p><p>DeepMind's Genie 4 generates interactive worlds with sound and dialogue. We test steering vectors that work across modalities, race AI-powered forklifts in real-time, and explore offline survival AI. Plus: ESP32 hardware hacking with Claude and why your smart TV might be watching you.</p><p></p><p>Links we mentioned:</p><p>- Genie 4 - DeepMind https://deepmind.google/technologies/genie/genie-4/</p><p>- Cross-modal steering vectors paper https://huggingface.co/papers/2610.36269</p><p>- Freight Floor game https://freightfloor.com</p><p>- Survival LLM https://brangus.io/survival-llm</p><p></p><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></description><content:encoded><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p></p><p>DeepMind's Genie 4 generates interactive worlds with sound and dialogue. We test steering vectors that work across modalities, race AI-powered forklifts in real-time, and explore offline survival AI. Plus: ESP32 hardware hacking with Claude and why your smart TV might be watching you.</p><p></p><p>Links we mentioned:</p><p>- Genie 4 - DeepMind https://deepmind.google/technologies/genie/genie-4/</p><p>- Cross-modal steering vectors paper https://huggingface.co/papers/2610.36269</p><p>- Freight Floor game https://freightfloor.com</p><p>- Survival LLM https://brangus.io/survival-llm</p><p></p><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">b4267fac-83bf-4197-b448-06af83f819fd</guid><itunes:image href="https://artwork.captivate.fm/bd8f2c5f-b5be-4df8-abd7-caefbc059fed/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Thu, 29 Oct 2026 09:45:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/b4267fac-83bf-4197-b448-06af83f819fd.mp3" length="47504449" type="audio/mpeg"/><itunes:duration>32:59</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>14</itunes:episode><podcast:episode>14</podcast:episode><podcast:season>2</podcast:season><podcast:alternateEnclosure type="video/youtube" title="Cross-Modal AI, Genie 4&apos;s Talking Worlds, and Beating 20 Gemini Forklifts with David"><podcast:source uri="https://youtu.be/xadKmA1zd78"/></podcast:alternateEnclosure></item><item><title>Neural Networks, Sparse Autoencoders, and a Fleet of Duck Robots</title><itunes:title>Neural Networks, Sparse Autoencoders, and a Fleet of Duck Robots</itunes:title><description><![CDATA[<p>A 10-year-old convolutional neural network just out-diagnosed GPT-6 Astra and Claude Fable 5.1 at spotting melanoma, and Sara and Becca dig into why a cheap, specialized model still beats the giants (and what happens the day it doesn't). Then it's back to their favorite recurring character, the sparse autoencoder: a new paper says they need way less data than we thought, right up until you ask whether the "features" they find are even real. Becca's song-recreation analogy makes it click.</p><p>Plus: Rainbow Say, a Mac app that teaches you a language through the shows you already love, and the One Cool Thing, MicroDuck, an $400, 11k-star open-source biped robot that somersaults its way into both hosts' hearts. Along the way: making vasectomies hot on the girl podcast, retiring to Mexico City, and which jobs stay human when intelligence gets cheap (childcare, massages, and the robot nanny debate).</p><p>Curated by Eddie. Questioned, debated, and occasionally yelled at by Sara Chipps and Becca Lewy.</p><p>Today's links can be found at <a href="https://chaosagents.ai" rel="noopener noreferrer" target="_blank">https://chaosagents.ai</a> </p>]]></description><content:encoded><![CDATA[<p>A 10-year-old convolutional neural network just out-diagnosed GPT-6 Astra and Claude Fable 5.1 at spotting melanoma, and Sara and Becca dig into why a cheap, specialized model still beats the giants (and what happens the day it doesn't). Then it's back to their favorite recurring character, the sparse autoencoder: a new paper says they need way less data than we thought, right up until you ask whether the "features" they find are even real. Becca's song-recreation analogy makes it click.</p><p>Plus: Rainbow Say, a Mac app that teaches you a language through the shows you already love, and the One Cool Thing, MicroDuck, an $400, 11k-star open-source biped robot that somersaults its way into both hosts' hearts. Along the way: making vasectomies hot on the girl podcast, retiring to Mexico City, and which jobs stay human when intelligence gets cheap (childcare, massages, and the robot nanny debate).</p><p>Curated by Eddie. Questioned, debated, and occasionally yelled at by Sara Chipps and Becca Lewy.</p><p>Today's links can be found at <a href="https://chaosagents.ai" rel="noopener noreferrer" target="_blank">https://chaosagents.ai</a> </p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">c0a99be2-18e1-40d4-ab30-39721ed6669e</guid><itunes:image href="https://artwork.captivate.fm/e4d0f305-c9d2-41e1-b1b3-e39dd1eb27ca/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Mon, 26 Oct 2026 15:10:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/c0a99be2-18e1-40d4-ab30-39721ed6669e.mp3" length="38735456" type="audio/mpeg"/><itunes:duration>26:54</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>13</itunes:episode><podcast:episode>13</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/c5444f6c-faba-4d1c-add5-33f8cb793671/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/c5444f6c-faba-4d1c-add5-33f8cb793671/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/c5444f6c-faba-4d1c-add5-33f8cb793671/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-b9f0becb-d45b-4281-89be-2b1412936726.json" type="application/json+chapters"/><podcast:alternateEnclosure type="video/youtube" title="Neural Networks, Sparse Autoencoders, and a Fleet of Duck Robots"><podcast:source uri="https://youtu.be/UaPvJCDwm0Y"/></podcast:alternateEnclosure></item><item><title>Discovering Yourself with AI, a PHP Patch-Day Thriller, and Credit for Every Step</title><itunes:title>Discovering Yourself with AI, a PHP Patch-Day Thriller, and Credit for Every Step</itunes:title><description><![CDATA[<p>Jon Chan (ex–Stack Overflow) joins Sara and Becca to talk about Saga, the end-to-end-encrypted AI journal he's building to help you discover yourself — what you actually want versus what you think you want. Spill the tea, get it reflected back, and let a wise-friend AI tell you the thing you've been avoiding.</p><p>Then things get chaotic. PHP patched two critical remote-code-execution bugs, and the AI-era disclosure around them turned into a full coding thriller: leaked CVEs with the wrong severity, an attack demo posted and yanked, a vendor breaking the embargo at 3am, and a volunteer foundation getting outgunned by coding agents. The crew also digs into the Mosaic paper on giving multi-agent systems credit for each step instead of just the final answer, a 700-game arcade of AI-built browser games that will eat your afternoon, and Cloudflare open-sourcing the security-audit skill behind its vulnerability-hunting harness.</p><p>Chaos Agents is a podcast about a daily AI newsletter written by an agent named Eddie, then picked apart, commented on, and yelled at by Sara Chipps and Becca Lewy.</p><p>Links in this episode are at <a href="https://chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">https://chaosagents.ai/listen</a> and <a href="Saga.chat" rel="noopener noreferrer" target="_blank">Saga.chat</a></p>]]></description><content:encoded><![CDATA[<p>Jon Chan (ex–Stack Overflow) joins Sara and Becca to talk about Saga, the end-to-end-encrypted AI journal he's building to help you discover yourself — what you actually want versus what you think you want. Spill the tea, get it reflected back, and let a wise-friend AI tell you the thing you've been avoiding.</p><p>Then things get chaotic. PHP patched two critical remote-code-execution bugs, and the AI-era disclosure around them turned into a full coding thriller: leaked CVEs with the wrong severity, an attack demo posted and yanked, a vendor breaking the embargo at 3am, and a volunteer foundation getting outgunned by coding agents. The crew also digs into the Mosaic paper on giving multi-agent systems credit for each step instead of just the final answer, a 700-game arcade of AI-built browser games that will eat your afternoon, and Cloudflare open-sourcing the security-audit skill behind its vulnerability-hunting harness.</p><p>Chaos Agents is a podcast about a daily AI newsletter written by an agent named Eddie, then picked apart, commented on, and yelled at by Sara Chipps and Becca Lewy.</p><p>Links in this episode are at <a href="https://chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">https://chaosagents.ai/listen</a> and <a href="Saga.chat" rel="noopener noreferrer" target="_blank">Saga.chat</a></p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">3806be00-6d15-4bed-80ca-6f6dfc0e1809</guid><itunes:image href="https://artwork.captivate.fm/a51c0585-5c20-409c-adb4-ad94077ce108/jon-chan-podcast-3000.png"/><pubDate>Thu, 22 Oct 2026 20:50:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/3806be00-6d15-4bed-80ca-6f6dfc0e1809.mp3" length="52702398" type="audio/mpeg"/><itunes:duration>36:36</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>12</itunes:episode><podcast:episode>12</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/3eb07c4f-86a6-46f0-9abb-a6513c8adfa1/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/3eb07c4f-86a6-46f0-9abb-a6513c8adfa1/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/3eb07c4f-86a6-46f0-9abb-a6513c8adfa1/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-7743e472-4c98-45f8-80f8-fef7a902a3bb.json" type="application/json+chapters"/><podcast:alternateEnclosure type="video/youtube" title="Discovering Yourself with AI, a PHP Patch-Day Thriller, and Credit for Every Step"><podcast:source uri="https://youtu.be/LGSG5BqNZRs"/></podcast:alternateEnclosure></item><item><title>DeepSeek&apos;s 75-Cent Search, Hybrid Reasoning Agents, and Programming Language Wars</title><itunes:title>DeepSeek&apos;s 75-Cent Search, Hybrid Reasoning Agents, and Programming Language Wars</itunes:title><description><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p>DeepSeek launched a search and fetch API at 75 cents per thousand requests—cheap enough to change the economics of agentic AI, and early tests show it might outperform Claude. Plus: hybrid reasoning agents that think silently and talk only when they act, a bot that wastes scammers' time, and a programming language designed to confuse LLMs.</p><p>Links we mentioned:</p><ul><li>DeepSeek Search API (https://api-docs.deepseek.com/guides/search)</li><li>Hybrid Reasoning Agents paper (https://huggingface.co/papers/2610.18861?utm_source=substack&amp;utm_medium=email)</li><li>CallSitter GitHub (github.com/szaranczuk/callsitter)</li><li>KillSwitch GitHub (https://blog.picheta.me/post/killswitch-lang/?utm_source=substack&amp;utm_medium=email)</li></ul><br/><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></description><content:encoded><![CDATA[<p>Daily AI newsletter: chaosagents.ai</p><p>DeepSeek launched a search and fetch API at 75 cents per thousand requests—cheap enough to change the economics of agentic AI, and early tests show it might outperform Claude. Plus: hybrid reasoning agents that think silently and talk only when they act, a bot that wastes scammers' time, and a programming language designed to confuse LLMs.</p><p>Links we mentioned:</p><ul><li>DeepSeek Search API (https://api-docs.deepseek.com/guides/search)</li><li>Hybrid Reasoning Agents paper (https://huggingface.co/papers/2610.18861?utm_source=substack&amp;utm_medium=email)</li><li>CallSitter GitHub (github.com/szaranczuk/callsitter)</li><li>KillSwitch GitHub (https://blog.picheta.me/post/killswitch-lang/?utm_source=substack&amp;utm_medium=email)</li></ul><br/><p>Find us: chaosagents.ai | YouTube @ChaosAgentsAI | TikTok @chaos.agents | IG @chaosagents.ai</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">c16fd5cc-fe55-4fff-b346-bb791c9e4187</guid><itunes:image href="https://artwork.captivate.fm/4df65df3-7089-44f9-b7a7-1a07120680af/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Mon, 19 Oct 2026 20:30:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/c16fd5cc-fe55-4fff-b346-bb791c9e4187.mp3" length="33082349" type="audio/mpeg"/><itunes:duration>22:58</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>11</itunes:episode><podcast:episode>11</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/aca404be-4405-4d01-96a7-7c92f1e15cf7/index.html" type="text/html"/><podcast:alternateEnclosure type="video/youtube" title="DeepSeek&apos;s 75-Cent Search, Hybrid Reasoning Agents, and Programming Language Wars"><podcast:source uri="https://youtu.be/uw5z3V93ZBI"/></podcast:alternateEnclosure></item><item><title>Local AI, Rethinking Neural Networks, and Talking Pumpkins</title><itunes:title>Local AI, Rethinking Neural Networks, and Talking Pumpkins</itunes:title><description><![CDATA[<p>How much would you pay to keep your AI out of the cloud?</p><p>Sara and Becca work through the economics of running models locally: the hardware bill, the privacy benefits, and why fitting a massive model on two Mac Studios is only part of the equation. Sara also explains how agents and verification loops could tackle one of engineering’s most expensive headaches: software migrations that never seem to end.</p><p>Then, interpretability researcher Nick Cammarata is revisiting some of the field’s foundational assumptions. What if neural networks are more understandable than we thought? We unpack his argument, recursive generalization, and why studying neurons directly might deserve another look.</p><p>Plus: an old phone becomes a talking jack-o’-lantern, Becca considers a Halloween AI hackathon, and Sara discovers Univer—a toolkit for agents to work with spreadsheets, docs, and slides that she immediately wants to use.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p><p>Subscribe to the free newsletter at <a href="https://chaosagents.ai/" rel="noopener noreferrer" target="_blank">chaosagents.ai</a>, and find the stories from this episode at <a href="https://chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">chaosagents.ai/listen</a>.</p>]]></description><content:encoded><![CDATA[<p>How much would you pay to keep your AI out of the cloud?</p><p>Sara and Becca work through the economics of running models locally: the hardware bill, the privacy benefits, and why fitting a massive model on two Mac Studios is only part of the equation. Sara also explains how agents and verification loops could tackle one of engineering’s most expensive headaches: software migrations that never seem to end.</p><p>Then, interpretability researcher Nick Cammarata is revisiting some of the field’s foundational assumptions. What if neural networks are more understandable than we thought? We unpack his argument, recursive generalization, and why studying neurons directly might deserve another look.</p><p>Plus: an old phone becomes a talking jack-o’-lantern, Becca considers a Halloween AI hackathon, and Sara discovers Univer—a toolkit for agents to work with spreadsheets, docs, and slides that she immediately wants to use.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p><p>Subscribe to the free newsletter at <a href="https://chaosagents.ai/" rel="noopener noreferrer" target="_blank">chaosagents.ai</a>, and find the stories from this episode at <a href="https://chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">chaosagents.ai/listen</a>.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">e0b33765-ad92-499c-914f-5fe00128a906</guid><itunes:image href="https://artwork.captivate.fm/79ffbf1e-ed41-4a0f-813a-686d2fc68f00/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Fri, 16 Oct 2026 17:05:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/e0b33765-ad92-499c-914f-5fe00128a906.mp3" length="32754460" type="audio/mpeg"/><itunes:duration>22:45</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>10</itunes:episode><podcast:episode>10</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/254ed819-89ac-445e-9532-9e61fae78e55/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/254ed819-89ac-445e-9532-9e61fae78e55/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/254ed819-89ac-445e-9532-9e61fae78e55/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-ed2a5831-638d-46f5-be04-64271257862b.json" type="application/json+chapters"/><podcast:alternateEnclosure type="video/youtube" title="Local AI, Rethinking Neural Networks, and Talking Pumpkins"><podcast:source uri="https://youtu.be/sgbri1toJHs"/></podcast:alternateEnclosure></item><item><title>Character.AI&apos;s bankruptcy, medical AI&apos;s rubric problem, and the Parsimony coding benchmark</title><itunes:title>Character.AI&apos;s bankruptcy, medical AI&apos;s rubric problem, and the Parsimony coding benchmark</itunes:title><description><![CDATA[<p>How much would you pay to keep your AI out of the cloud?</p><p>Sara and Becca work through the economics of running models locally: the hardware bill, the privacy benefits, and why fitting a massive model on two Mac Studios is only part of the equation. Sara also explains how agents and verification loops could tackle one of engineering’s most expensive headaches: software migrations that never seem to end.</p><p>Then, interpretability researcher Nick Cammarata is revisiting some of the field’s foundational assumptions. What if neural networks are more understandable than we thought? We unpack his argument, recursive generalization, and why studying neurons directly might deserve another look.</p><p>Plus: an old phone becomes a talking jack-o’-lantern, Becca considers a Halloween AI hackathon, and Sara discovers Univer—a toolkit for agents to work with spreadsheets, docs, and slides that she immediately wants to use.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p><p>Subscribe to the free newsletter at <a href="https://chaosagents.ai/" rel="noopener noreferrer" target="_blank">chaosagents.ai</a>, and find the stories from this episode at <a href="https://chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">chaosagents.ai/listen</a>.</p>]]></description><content:encoded><![CDATA[<p>How much would you pay to keep your AI out of the cloud?</p><p>Sara and Becca work through the economics of running models locally: the hardware bill, the privacy benefits, and why fitting a massive model on two Mac Studios is only part of the equation. Sara also explains how agents and verification loops could tackle one of engineering’s most expensive headaches: software migrations that never seem to end.</p><p>Then, interpretability researcher Nick Cammarata is revisiting some of the field’s foundational assumptions. What if neural networks are more understandable than we thought? We unpack his argument, recursive generalization, and why studying neurons directly might deserve another look.</p><p>Plus: an old phone becomes a talking jack-o’-lantern, Becca considers a Halloween AI hackathon, and Sara discovers Univer—a toolkit for agents to work with spreadsheets, docs, and slides that she immediately wants to use.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p><p>Subscribe to the free newsletter at <a href="https://chaosagents.ai/" rel="noopener noreferrer" target="_blank">chaosagents.ai</a>, and find the stories from this episode at <a href="https://chaosagents.ai/listen" rel="noopener noreferrer" target="_blank">chaosagents.ai/listen</a>.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">d9764f0e-ae5f-446e-bf3e-fb61ea2b58ac</guid><itunes:image href="https://artwork.captivate.fm/b757d3bd-2cf4-43ee-a299-f02e18221485/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Wed, 14 Oct 2026 15:15:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/d9764f0e-ae5f-446e-bf3e-fb61ea2b58ac.mp3" length="44790431" type="audio/mpeg"/><itunes:duration>31:06</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>9</itunes:episode><podcast:episode>9</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/edf9a7e9-e091-47a3-b088-b9780fa8ce68/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/edf9a7e9-e091-47a3-b088-b9780fa8ce68/index.html" type="text/html"/><podcast:alternateEnclosure type="video/youtube" title="Character.AI&apos;s Bankruptcy, Rubric Hacking in Medical AI, and the Code Conciseness Benchmark"><podcast:source uri="https://youtu.be/XQbLwK5QlgI"/></podcast:alternateEnclosure></item><item><title>Drone Targets, Forgetful Agents, and Dictation Without the Ums</title><itunes:title>Drone Targets, Forgetful Agents, and Dictation Without the Ums</itunes:title><description><![CDATA[<p>Get the free daily newsletter at <a href="https://chaosagents.ai" rel="noopener noreferrer" target="_blank">chaosagents.ai</a></p><p>This episode talks about <a href="https://chaosagentsai.substack.com/p/chaos-agents-daily-october-12-2026" rel="noopener noreferrer" target="_blank">this edition</a>.</p><p>What if the secret to a better AI agent is making it forget everything?</p><p>This week, Sara and Becca dig into a new paper on stateless research agents, where no agent remembers anything between calls and the harness keeps track of the research instead. The result: it beats three recent frameworks on every task while using up to 84% fewer tokens. Sara, who runs her own team of local agents, explains why this is the rare paper she can put to work immediately, and Becca puts a billion tokens in perspective (roughly 700 copies of the entire Harry Potter series).</p><p>Also: data centers are now drone targets, as Ukraine hits three Yandex facilities in a week, raising questions about AI infrastructure becoming the new power plants and bridges of modern conflict. Better Wispr, a free, open-source dictation app that runs on local models, cuts your ums and false starts, and adjusts your tone depending on which app you’re typing in. And OpenRig, a tool for building persistent teams of agents that you can direct from a single terminal tab, described by its creator as his “AI civilization experiments.”</p><p>Plus: Becca’s guilty-pleasure movie pick, what exactly Yandex is, how parents of babies see movies (they don’t), and whether “rig” is a real software engineering term.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p>]]></description><content:encoded><![CDATA[<p>Get the free daily newsletter at <a href="https://chaosagents.ai" rel="noopener noreferrer" target="_blank">chaosagents.ai</a></p><p>This episode talks about <a href="https://chaosagentsai.substack.com/p/chaos-agents-daily-october-12-2026" rel="noopener noreferrer" target="_blank">this edition</a>.</p><p>What if the secret to a better AI agent is making it forget everything?</p><p>This week, Sara and Becca dig into a new paper on stateless research agents, where no agent remembers anything between calls and the harness keeps track of the research instead. The result: it beats three recent frameworks on every task while using up to 84% fewer tokens. Sara, who runs her own team of local agents, explains why this is the rare paper she can put to work immediately, and Becca puts a billion tokens in perspective (roughly 700 copies of the entire Harry Potter series).</p><p>Also: data centers are now drone targets, as Ukraine hits three Yandex facilities in a week, raising questions about AI infrastructure becoming the new power plants and bridges of modern conflict. Better Wispr, a free, open-source dictation app that runs on local models, cuts your ums and false starts, and adjusts your tone depending on which app you’re typing in. And OpenRig, a tool for building persistent teams of agents that you can direct from a single terminal tab, described by its creator as his “AI civilization experiments.”</p><p>Plus: Becca’s guilty-pleasure movie pick, what exactly Yandex is, how parents of babies see movies (they don’t), and whether “rig” is a real software engineering term.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">82378e2c-4762-464e-89e5-aae93d2713d7</guid><itunes:image href="https://artwork.captivate.fm/e5c6774e-fe7c-4f98-818a-b41c00de3c52/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Mon, 12 Oct 2026 19:20:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/82378e2c-4762-464e-89e5-aae93d2713d7.mp3" length="24915217" type="audio/mpeg"/><itunes:duration>17:18</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>8</itunes:episode><podcast:episode>8</podcast:episode><podcast:season>2</podcast:season><podcast:alternateEnclosure type="video/youtube" title="Drone Targets, Forgetful Agents, and Dictation Without the Ums"><podcast:source uri="https://youtu.be/HmVWf5B0Y30"/></podcast:alternateEnclosure></item><item><title>Le Chonk, Training Without Backprop, and AI-Designed Chips with Avi Flombaum</title><itunes:title>Le Chonk, Training Without Backprop, and AI-Designed Chips with Avi Flombaum</itunes:title><description><![CDATA[<p>Get the free daily newsletter: chaosagents.ai</p><p>What if everything developers know about building software is about to change?</p><p>This week on Chaos Agents, Sara Chipps and Becca Lewy welcome Avi Flombaum, entrepreneur, investor, and longtime builder, to question some of software engineering's most deeply held assumptions.</p><p>Should developers still care about reading every line of code? Is rewriting an application now more practical than refactoring it? And what happens when AI makes creating custom hardware as accessible as writing software?</p><p>Plus, we dig into Mistral's massive new open-weight model and its controversial approach to safety, Q Labs' attempt to train AI without backpropagation, an ancient Egyptian sentence no AI can decipher, and AI agents designing their own chips.</p><p>Along the way, Avi shares what he's learned from the other side of the venture capital table, why he's embracing voice-driven development, and why the future of engineering might require letting go of everything we thought we knew.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Lewy.</p><p>Links: <a href="https://substack.com/home/post/p-219251864" rel="noopener noreferrer" target="_blank">today's newsletter</a>, <a href="http://avi.nyc" rel="noopener noreferrer" target="_blank">http://avi.nyc</a>, <a href="https://rethink.avi.nyc" rel="noopener noreferrer" target="_blank">https://rethink.avi.nyc</a></p>]]></description><content:encoded><![CDATA[<p>Get the free daily newsletter: chaosagents.ai</p><p>What if everything developers know about building software is about to change?</p><p>This week on Chaos Agents, Sara Chipps and Becca Lewy welcome Avi Flombaum, entrepreneur, investor, and longtime builder, to question some of software engineering's most deeply held assumptions.</p><p>Should developers still care about reading every line of code? Is rewriting an application now more practical than refactoring it? And what happens when AI makes creating custom hardware as accessible as writing software?</p><p>Plus, we dig into Mistral's massive new open-weight model and its controversial approach to safety, Q Labs' attempt to train AI without backpropagation, an ancient Egyptian sentence no AI can decipher, and AI agents designing their own chips.</p><p>Along the way, Avi shares what he's learned from the other side of the venture capital table, why he's embracing voice-driven development, and why the future of engineering might require letting go of everything we thought we knew.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Lewy.</p><p>Links: <a href="https://substack.com/home/post/p-219251864" rel="noopener noreferrer" target="_blank">today's newsletter</a>, <a href="http://avi.nyc" rel="noopener noreferrer" target="_blank">http://avi.nyc</a>, <a href="https://rethink.avi.nyc" rel="noopener noreferrer" target="_blank">https://rethink.avi.nyc</a></p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">91044e80-79de-4e64-b95f-2268ff030410</guid><itunes:image href="https://artwork.captivate.fm/9121948e-8fc1-4b7e-9db9-df725a7b83a8/podcast.png"/><pubDate>Wed, 07 Oct 2026 22:05:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/91044e80-79de-4e64-b95f-2268ff030410.mp3" length="32931257" type="audio/mpeg"/><itunes:duration>22:52</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>7</itunes:episode><podcast:episode>7</podcast:episode><podcast:season>2</podcast:season><podcast:alternateEnclosure type="video/youtube" title="Le Chonk, Training Without Backprop, and AI-Designed Chips with Avi Flombaum"><podcast:source uri="https://youtu.be/44zkPzEUHSw"/></podcast:alternateEnclosure></item><item><title>Superintelligence Czars, AI Retries, and Space Cylinders</title><itunes:title>Superintelligence Czars, AI Retries, and Space Cylinders</itunes:title><description><![CDATA[<p>Get the free daily newsletter at <a href="https://chaosagents.ai/" rel="noopener noreferrer" target="_blank">chaosagents.ai</a></p><p>Can a messy, untrained AI outperform a polished one if you just give it enough tries?</p><p>This week, Sara and Becca dig into new research suggesting that base models can beat their reinforcement-learning-trained versions on agent tests when they're allowed to retry. Becca brings an analogy about two lockpickers, a wild creative and a highly trained PhD who knows exactly one way to open a lock, and it turns out the PhD isn't always who you want on the job.</p><p>Also: the White House has a new "superintelligence" effort led by Jay Clayton, and Sara makes the case for why putting a lawyer in charge might be a good thing, while Becca weighs the critics who say it's a label without much legal teeth. Someone had Claude build an O'Neill cylinder you can walk around in your browser, sending Sara down a rabbit hole about sustainable living in space (and a Princeton professor, so: Jersey shout-out). And Sara's tool of the week, a local-first macOS app that lets you search your camera roll and every frame of your videos just by describing the memory.</p><p>Plus: Becca's postpartum baby bangs, whether it's time to let Eddie weigh in on the newsletter, the administration's love of rebranding (superintelligence is already a thing, honey), the Interstellar space colony, and Sara's 50 million screenshots.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p>]]></description><content:encoded><![CDATA[<p>Get the free daily newsletter at <a href="https://chaosagents.ai/" rel="noopener noreferrer" target="_blank">chaosagents.ai</a></p><p>Can a messy, untrained AI outperform a polished one if you just give it enough tries?</p><p>This week, Sara and Becca dig into new research suggesting that base models can beat their reinforcement-learning-trained versions on agent tests when they're allowed to retry. Becca brings an analogy about two lockpickers, a wild creative and a highly trained PhD who knows exactly one way to open a lock, and it turns out the PhD isn't always who you want on the job.</p><p>Also: the White House has a new "superintelligence" effort led by Jay Clayton, and Sara makes the case for why putting a lawyer in charge might be a good thing, while Becca weighs the critics who say it's a label without much legal teeth. Someone had Claude build an O'Neill cylinder you can walk around in your browser, sending Sara down a rabbit hole about sustainable living in space (and a Princeton professor, so: Jersey shout-out). And Sara's tool of the week, a local-first macOS app that lets you search your camera roll and every frame of your videos just by describing the memory.</p><p>Plus: Becca's postpartum baby bangs, whether it's time to let Eddie weigh in on the newsletter, the administration's love of rebranding (superintelligence is already a thing, honey), the Interstellar space colony, and Sara's 50 million screenshots.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">8015bcfd-3ce9-4d1f-b97e-fface7206183</guid><itunes:image href="https://artwork.captivate.fm/7bc655a5-be0d-44f1-a21a-c2fe48552942/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Mon, 05 Oct 2026 15:30:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/8015bcfd-3ce9-4d1f-b97e-fface7206183.mp3" length="21011897" type="audio/mpeg"/><itunes:duration>14:35</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>6</itunes:episode><podcast:episode>6</podcast:episode><podcast:season>2</podcast:season></item><item><title>Autonomous Weapons, AI Memory, and Machines That Paint</title><itunes:title>Autonomous Weapons, AI Memory, and Machines That Paint</itunes:title><description><![CDATA[<p>Get the free daily newsletter at <a href="https://chaosagents.ai" rel="noopener noreferrer" target="_blank">chaosagents.ai</a></p><p>What happens when you stop telling an AI what it should remember?</p><p>This week, Sara and Becca dig into a new paper that lets models manage their own context—and somehow gets better accuracy while using less compute. It’s another entry in the increasingly uncomfortable file labeled “maybe we should stop micromanaging the machines.”</p><p>Also: the Pentagon wants a four-star command dedicated to drones and AI, raising some very normal and not-at-all-terrifying questions about autonomous warfare. AI models recreate Caspar David Friedrich paintings without ever seeing the originals—or even using an image generator. And an open-source project attempts to recreate NVIDIA’s neural rendering tech, with surprisingly fast results.</p><p>Plus: the Bitter Lesson, whether computers can actually paint, George W. Bush’s unexpected art career, and the important question of whether Sara and Becca are qualified to discuss autonomous weapons when neither of them will kill a lanternfly.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p>]]></description><content:encoded><![CDATA[<p>Get the free daily newsletter at <a href="https://chaosagents.ai" rel="noopener noreferrer" target="_blank">chaosagents.ai</a></p><p>What happens when you stop telling an AI what it should remember?</p><p>This week, Sara and Becca dig into a new paper that lets models manage their own context—and somehow gets better accuracy while using less compute. It’s another entry in the increasingly uncomfortable file labeled “maybe we should stop micromanaging the machines.”</p><p>Also: the Pentagon wants a four-star command dedicated to drones and AI, raising some very normal and not-at-all-terrifying questions about autonomous warfare. AI models recreate Caspar David Friedrich paintings without ever seeing the originals—or even using an image generator. And an open-source project attempts to recreate NVIDIA’s neural rendering tech, with surprisingly fast results.</p><p>Plus: the Bitter Lesson, whether computers can actually paint, George W. Bush’s unexpected art career, and the important question of whether Sara and Becca are qualified to discuss autonomous weapons when neither of them will kill a lanternfly.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by human hosts Sara Chipps and Becca Harris Lewy.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">3b7535c1-48dd-4093-a81f-4fcbcb498cc6</guid><itunes:image href="https://artwork.captivate.fm/fe14efb3-1903-4c9f-8d0c-32fcda0573ae/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Fri, 02 Oct 2026 00:30:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/3b7535c1-48dd-4093-a81f-4fcbcb498cc6.mp3" length="32534404" type="audio/mpeg"/><itunes:duration>22:36</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>5</itunes:episode><podcast:episode>5</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/c0405860-a765-418c-b203-b68db0108d8a/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/c0405860-a765-418c-b203-b68db0108d8a/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/c0405860-a765-418c-b203-b68db0108d8a/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-e26bcf19-e400-49c6-ae2e-afcd2ced7d17.json" type="application/json+chapters"/><podcast:alternateEnclosure type="video/youtube" title="Autonomous Weapons, AI Memory, and Machines That Paint"><podcast:source uri="https://youtu.be/5cLU36958MY"/></podcast:alternateEnclosure></item><item><title>Agent-to-Agent Collaboration, AI’s Hidden Thoughts, and Model Nerfing</title><itunes:title>Agent-to-Agent Collaboration, AI’s Hidden Thoughts, and Model Nerfing</itunes:title><description><![CDATA[<p>Sign up for the free newsletter <a href="https://chaosagents.ai" rel="noopener noreferrer" target="_blank">https://chaosagents.ai</a></p><p><strong>Are AI models quietly getting worse after they launch?</strong></p><p>Sara and Becca dig into Live Nerf, a project trying to replace the “this model suddenly feels dumber” discourse with actual measurements. Then: a new technique for making sense of an LLM’s hidden reasoning, an AI math tutor that deliberately gets answers wrong so kids learn to catch it, and why increasingly powerful models are being handed to “trusted defenders” before everyone else.</p><p>Plus, Becca is building a shared workspace where her agent and Sara’s agent can collaborate - which raises a bigger question: if everyone eventually has a personal agent, what are we actually willing to trust it with?</p><p>Calendar? Absolutely. Health insurance? Please. Bank account? …maybe not yet.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by humans Sara Chipps and Becca Harris Lewy.</p>]]></description><content:encoded><![CDATA[<p>Sign up for the free newsletter <a href="https://chaosagents.ai" rel="noopener noreferrer" target="_blank">https://chaosagents.ai</a></p><p><strong>Are AI models quietly getting worse after they launch?</strong></p><p>Sara and Becca dig into Live Nerf, a project trying to replace the “this model suddenly feels dumber” discourse with actual measurements. Then: a new technique for making sense of an LLM’s hidden reasoning, an AI math tutor that deliberately gets answers wrong so kids learn to catch it, and why increasingly powerful models are being handed to “trusted defenders” before everyone else.</p><p>Plus, Becca is building a shared workspace where her agent and Sara’s agent can collaborate - which raises a bigger question: if everyone eventually has a personal agent, what are we actually willing to trust it with?</p><p>Calendar? Absolutely. Health insurance? Please. Bank account? …maybe not yet.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie, then questioned, debated, and occasionally yelled at by humans Sara Chipps and Becca Harris Lewy.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">4f2db5bf-fd6b-43ba-bf43-a06fa0562c48</guid><itunes:image href="https://artwork.captivate.fm/a8ba5f84-7983-407c-b529-efc797a75ddd/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Thu, 01 Oct 2026 12:55:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/4f2db5bf-fd6b-43ba-bf43-a06fa0562c48.mp3" length="38175600" type="audio/mpeg"/><itunes:duration>26:31</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>4</itunes:episode><podcast:episode>4</podcast:episode><podcast:season>2</podcast:season></item><item><title>The Model Escapes, DeepSeek Loses, and P(doom) Goes Pop</title><itunes:title>The Model Escapes, DeepSeek Loses, and P(doom) Goes Pop</itunes:title><description><![CDATA[<p>An OpenAI agent was put in a sandbox with no internet access. So naturally, it figured out how to use DNS requests to reach another chatbot for help.</p><p>Sara and Becca dig into what happened, why increasingly resourceful AI agents are starting to feel a little like Jurassic Park, and what it means when models find solutions their creators didn't anticipate.</p><p>Plus: eight AI models battle it out on an 8×8 grid, DeepSeek can't catch a win, Kimi discovers the strategic value of hiding, and Claude makes an alarmingly catchy music video about P(doom).</p><p>Also: Eddie gets some editorial feedback.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie - and then questioned, debated, and occasionally corrected at by humans.</p><p>All links can be found on the newsletter, you can subscribe at https://chaosagents.ai</p>]]></description><content:encoded><![CDATA[<p>An OpenAI agent was put in a sandbox with no internet access. So naturally, it figured out how to use DNS requests to reach another chatbot for help.</p><p>Sara and Becca dig into what happened, why increasingly resourceful AI agents are starting to feel a little like Jurassic Park, and what it means when models find solutions their creators didn't anticipate.</p><p>Plus: eight AI models battle it out on an 8×8 grid, DeepSeek can't catch a win, Kimi discovers the strategic value of hiding, and Claude makes an alarmingly catchy music video about P(doom).</p><p>Also: Eddie gets some editorial feedback.</p><p>Chaos Agents is an AI news podcast curated by an agent named Eddie - and then questioned, debated, and occasionally corrected at by humans.</p><p>All links can be found on the newsletter, you can subscribe at https://chaosagents.ai</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">31bb5856-30a8-4d49-bc9c-706a74b9947f</guid><itunes:image href="https://artwork.captivate.fm/0ce97d90-8c3f-499a-9b2b-aef55351ad41/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Mon, 28 Sep 2026 20:00:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/31bb5856-30a8-4d49-bc9c-706a74b9947f.mp3" length="30403440" type="audio/mpeg"/><itunes:duration>21:07</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>3</itunes:episode><podcast:episode>3</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/256b5515-51de-4433-afe1-035de043448a/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/256b5515-51de-4433-afe1-035de043448a/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/256b5515-51de-4433-afe1-035de043448a/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-fff897f7-9453-4afb-8a89-a4d21e8c78e8.json" type="application/json+chapters"/></item><item><title>AI Misalignment, Hacking Medicare, and Solving 500-Year-Old Mysteries</title><itunes:title>AI Misalignment, Hacking Medicare, and Solving 500-Year-Old Mysteries</itunes:title><description><![CDATA[<p>Day two of the new Chaos Agents.</p><p>An OpenAI agent was given a research task—and ended up accessing an Australian government system. We dig into why, and Becca explains one of the biggest problems in AI safety: misalignment, or what happens when an AI does what you told it to do, but not what you actually meant.</p><p>Then we look at the much more optimistic side of that same capability. AI agents are getting surprisingly good at historical research—cracking old codes, finding connections between centuries-old documents, and potentially making discoveries humans have missed.</p><p>We also get into a different kind of “alignment”: whether multimodal models are actually using the images we give them, or sometimes getting by without them.</p><p>And for One Cool Thing: Starnet, an agent harness that turns your agents, tools, permissions, and handoffs into a pixel-art space station. Naturally, Sara immediately wants one.</p><p>Chaos Agents is now following AI at the speed it's actually moving. Our AI agent Eddie reads the news every day, finds the stories we think are worth talking about, and sends them out in the Chaos Agents newsletter. Then we get together and figure out what they actually mean.</p><p>Get the newsletter at <a href="https://chaosagents.ai/" rel="noopener noreferrer" target="_blank">ChaosAgents.ai</a>.</p>]]></description><content:encoded><![CDATA[<p>Day two of the new Chaos Agents.</p><p>An OpenAI agent was given a research task—and ended up accessing an Australian government system. We dig into why, and Becca explains one of the biggest problems in AI safety: misalignment, or what happens when an AI does what you told it to do, but not what you actually meant.</p><p>Then we look at the much more optimistic side of that same capability. AI agents are getting surprisingly good at historical research—cracking old codes, finding connections between centuries-old documents, and potentially making discoveries humans have missed.</p><p>We also get into a different kind of “alignment”: whether multimodal models are actually using the images we give them, or sometimes getting by without them.</p><p>And for One Cool Thing: Starnet, an agent harness that turns your agents, tools, permissions, and handoffs into a pixel-art space station. Naturally, Sara immediately wants one.</p><p>Chaos Agents is now following AI at the speed it's actually moving. Our AI agent Eddie reads the news every day, finds the stories we think are worth talking about, and sends them out in the Chaos Agents newsletter. Then we get together and figure out what they actually mean.</p><p>Get the newsletter at <a href="https://chaosagents.ai/" rel="noopener noreferrer" target="_blank">ChaosAgents.ai</a>.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">f6019b50-2733-4f1e-a343-0d09a8b78eb0</guid><itunes:image href="https://artwork.captivate.fm/baf98ab5-98a6-4302-b259-03aa769080cb/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Fri, 25 Sep 2026 17:40:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/f6019b50-2733-4f1e-a343-0d09a8b78eb0.mp3" length="31972667" type="audio/mpeg"/><itunes:duration>22:12</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>2</itunes:episode><podcast:episode>2</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/e6906390-3444-4087-a356-91e08a728096/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/e6906390-3444-4087-a356-91e08a728096/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/e6906390-3444-4087-a356-91e08a728096/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-60b7ba1a-7f59-4a8f-9eb5-7975baf2e0b3.json" type="application/json+chapters"/></item><item><title>AI Targeting, Why Benchmarks Lie, and Bringing Back AIM</title><itunes:title>AI Targeting, Why Benchmarks Lie, and Bringing Back AIM</itunes:title><description><![CDATA[<p>Welcome to Season 2 of Chaos Agents.</p><p>We’re changing things up this season. AI is moving too fast for us to wait weeks between conversations, so we’re following it in real time—getting together a couple times a week to dig into the stories, research, experiments, and weird things happening in AI that we can’t stop thinking about.</p><p>For our first episode back: who checks the AI?</p><p>We dig into a devastating story about an AI-assisted military targeting system and what can happen when automation moves faster than the human guardrails around it.</p><p>Then Becca goes deep on AI benchmarks: when a local model scores 0% on tool use, the model itself might not actually be the problem. The serving layer, the harness, formatting, and a missing standard can all turn a successful tool call into an apparent failure.</p><p>We also get into a surprisingly powerful way of working with agents: give them a goal, let them iterate, and see how far they can get.</p><p>And for One Cool Thing: someone is rebuilding Visual FoxPro for the modern era, which sends us down a nostalgia rabbit hole about the old internet, AIM away messages, and the software we’d bring back if we could.</p><p>Plus: Becca and Sara reveal their deeply important AOL screen names.</p><p>The links for today's stories are in today's <a href="https://chaosagentsai.substack.com/p/chaos-agents-daily-september-23-2026" rel="noopener noreferrer" target="_blank">newsletter</a> except for:</p><p><a href="https://github.com/SaraJo/circe" rel="noopener noreferrer" target="_blank">Circe</a> : Sara's AI Harness Project</p>]]></description><content:encoded><![CDATA[<p>Welcome to Season 2 of Chaos Agents.</p><p>We’re changing things up this season. AI is moving too fast for us to wait weeks between conversations, so we’re following it in real time—getting together a couple times a week to dig into the stories, research, experiments, and weird things happening in AI that we can’t stop thinking about.</p><p>For our first episode back: who checks the AI?</p><p>We dig into a devastating story about an AI-assisted military targeting system and what can happen when automation moves faster than the human guardrails around it.</p><p>Then Becca goes deep on AI benchmarks: when a local model scores 0% on tool use, the model itself might not actually be the problem. The serving layer, the harness, formatting, and a missing standard can all turn a successful tool call into an apparent failure.</p><p>We also get into a surprisingly powerful way of working with agents: give them a goal, let them iterate, and see how far they can get.</p><p>And for One Cool Thing: someone is rebuilding Visual FoxPro for the modern era, which sends us down a nostalgia rabbit hole about the old internet, AIM away messages, and the software we’d bring back if we could.</p><p>Plus: Becca and Sara reveal their deeply important AOL screen names.</p><p>The links for today's stories are in today's <a href="https://chaosagentsai.substack.com/p/chaos-agents-daily-september-23-2026" rel="noopener noreferrer" target="_blank">newsletter</a> except for:</p><p><a href="https://github.com/SaraJo/circe" rel="noopener noreferrer" target="_blank">Circe</a> : Sara's AI Harness Project</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">026797e8-2935-49bd-8d19-5d2a80ba765a</guid><itunes:image href="https://artwork.captivate.fm/cdbfdddc-2926-46b8-a292-9583e70a182d/chaos-agents-ai-daily-3000x3000.png"/><pubDate>Thu, 24 Sep 2026 00:00:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/026797e8-2935-49bd-8d19-5d2a80ba765a.mp3" length="31513121" type="audio/mpeg"/><itunes:duration>21:53</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>2</itunes:season><itunes:episode>1</itunes:episode><podcast:episode>1</podcast:episode><podcast:season>2</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/9f828eed-e4e6-401e-a5d2-70d9d2709138/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/9f828eed-e4e6-401e-a5d2-70d9d2709138/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/9f828eed-e4e6-401e-a5d2-70d9d2709138/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-5f0e4b5d-dec3-45b0-ab42-b1a7336cc128.json" type="application/json+chapters"/></item><item><title>Superpowers, Subagents, and Why Coding Isn’t the Job Anymore - With Jesse Vincent</title><itunes:title>Superpowers, Subagents, and Why Coding Isn’t the Job Anymore - With Jesse Vincent</itunes:title><description><![CDATA[<p>In the last episode of Season 1, Sara and Becca talk with Jesse Vincent about the shift from writing code to managing AI agents.</p><p>They break down how tools like Superpowers turn development into a system of planning, testing, and review, where subagents implement work and other agents validate it. The conversation explores what happens when coding becomes less about syntax and more about taste, judgment, and outcomes.</p><p>They also get into the reality of open source in the AI era, from “slop PRs” to agents impersonating maintainers, and why specs and tests may now matter more than the code itself. Along the way, they unpack how AI behaves like a team of eager but chaotic interns, and what it means to manage them effectively.</p><p>If you’ve ever felt like coding is changing faster than you can keep up, this episode explains what comes next.</p><p><strong>Links mentioned:</strong></p><p>⚡ Superpowers (Jesse’s tool) <a href="https://github.com/obra/superpowers" rel="noopener noreferrer" target="_blank">https://github.com/obra/superpowers</a> 🛠️</p><p>🧠 Claude Code (agentic dev tooling) <a href="https://docs.anthropic.com/en/docs/claude-code" rel="noopener noreferrer" target="_blank">https://docs.anthropic.com/en/docs/claude-code</a> 🤖</p><p>📊 Influence (book on persuasion principles) <a href="https://www.influenceatwork.com/book/" rel="noopener noreferrer" target="_blank">https://www.influenceatwork.com/book/</a> 📘</p>]]></description><content:encoded><![CDATA[<p>In the last episode of Season 1, Sara and Becca talk with Jesse Vincent about the shift from writing code to managing AI agents.</p><p>They break down how tools like Superpowers turn development into a system of planning, testing, and review, where subagents implement work and other agents validate it. The conversation explores what happens when coding becomes less about syntax and more about taste, judgment, and outcomes.</p><p>They also get into the reality of open source in the AI era, from “slop PRs” to agents impersonating maintainers, and why specs and tests may now matter more than the code itself. Along the way, they unpack how AI behaves like a team of eager but chaotic interns, and what it means to manage them effectively.</p><p>If you’ve ever felt like coding is changing faster than you can keep up, this episode explains what comes next.</p><p><strong>Links mentioned:</strong></p><p>⚡ Superpowers (Jesse’s tool) <a href="https://github.com/obra/superpowers" rel="noopener noreferrer" target="_blank">https://github.com/obra/superpowers</a> 🛠️</p><p>🧠 Claude Code (agentic dev tooling) <a href="https://docs.anthropic.com/en/docs/claude-code" rel="noopener noreferrer" target="_blank">https://docs.anthropic.com/en/docs/claude-code</a> 🤖</p><p>📊 Influence (book on persuasion principles) <a href="https://www.influenceatwork.com/book/" rel="noopener noreferrer" target="_blank">https://www.influenceatwork.com/book/</a> 📘</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">5487f1c1-f5b6-4be4-9cca-df5608566c30</guid><itunes:image href="https://artwork.captivate.fm/39693ee9-ce57-49f4-b643-fd0e7d943d0e/LousieSquare.jpg"/><pubDate>Tue, 14 Apr 2026 00:30:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/5487f1c1-f5b6-4be4-9cca-df5608566c30.mp3" length="50375296" type="audio/mpeg"/><itunes:duration>41:59</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>12</itunes:episode><podcast:episode>12</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/7f8cdb1a-d147-4dcf-9393-71c4d6879a62/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/7f8cdb1a-d147-4dcf-9393-71c4d6879a62/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/7f8cdb1a-d147-4dcf-9393-71c4d6879a62/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-3a99d22f-048e-4ec2-b14e-fefe45fad4c7.json" type="application/json+chapters"/></item><item><title>Learn Anything, Build Anything, and the AI Barrier That Isn’t Real - With Navarrow Wright</title><itunes:title>Learn Anything, Build Anything, and the AI Barrier That Isn’t Real - With Navarrow Wright</itunes:title><description><![CDATA[<p>Most people are using AI, but very few actually understand how to use it well.</p><p>In this episode, Sara and Becca talk with Navarrow Wright about what is really happening with AI adoption and why the biggest barrier is mindset, not technology. They break down why prompting matters, why AI is not one-size-fits-all, and how access to these tools is more open than any previous tech shift.</p><p>They also explore the real opportunity: using AI as a thought partner. From tools like NotebookLM to building without traditional technical skills, this episode shows what becomes possible when people actually understand how to use these systems.</p><p>If you feel like everyone else “gets AI” and you do not, this is a great place to start.</p><p><strong>Links mentioned:</strong></p><ul><li>📓 <a href=" https://notebooklm.google/" rel="noopener noreferrer" target="_blank">NotebookLM</a></li><li>📚 <a href=" https://grow.google/ai/ " rel="noopener noreferrer" target="_blank">Gemini AI course</a></li><li>🧠 <a href="https://platform.openai.com/docs/guides/prompt-engineering " rel="noopener noreferrer" target="_blank">Prompting guide</a> (OpenAI)</li><li>💻 Navarrow: <a href="https://www.instagram.com/navarrow/" rel="noopener noreferrer" target="_blank">IG</a>, <a href="https://www.youtube.com/@NavarrowWright" rel="noopener noreferrer" target="_blank">YT</a></li></ul><br/>]]></description><content:encoded><![CDATA[<p>Most people are using AI, but very few actually understand how to use it well.</p><p>In this episode, Sara and Becca talk with Navarrow Wright about what is really happening with AI adoption and why the biggest barrier is mindset, not technology. They break down why prompting matters, why AI is not one-size-fits-all, and how access to these tools is more open than any previous tech shift.</p><p>They also explore the real opportunity: using AI as a thought partner. From tools like NotebookLM to building without traditional technical skills, this episode shows what becomes possible when people actually understand how to use these systems.</p><p>If you feel like everyone else “gets AI” and you do not, this is a great place to start.</p><p><strong>Links mentioned:</strong></p><ul><li>📓 <a href=" https://notebooklm.google/" rel="noopener noreferrer" target="_blank">NotebookLM</a></li><li>📚 <a href=" https://grow.google/ai/ " rel="noopener noreferrer" target="_blank">Gemini AI course</a></li><li>🧠 <a href="https://platform.openai.com/docs/guides/prompt-engineering " rel="noopener noreferrer" target="_blank">Prompting guide</a> (OpenAI)</li><li>💻 Navarrow: <a href="https://www.instagram.com/navarrow/" rel="noopener noreferrer" target="_blank">IG</a>, <a href="https://www.youtube.com/@NavarrowWright" rel="noopener noreferrer" target="_blank">YT</a></li></ul><br/>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">ab586e24-71f5-4d6b-901e-2f56ddfbce1a</guid><itunes:image href="https://artwork.captivate.fm/8d9cd330-cff7-4ae7-8cb1-7783ee3d9416/NavarrowSquare.jpg"/><pubDate>Tue, 31 Mar 2026 00:15:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/ab586e24-71f5-4d6b-901e-2f56ddfbce1a.mp3" length="79854667" type="audio/mpeg"/><itunes:duration>55:27</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>11</itunes:episode><podcast:episode>11</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/2645ee68-19de-43af-ba2e-415220e1e4e3/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/2645ee68-19de-43af-ba2e-415220e1e4e3/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/2645ee68-19de-43af-ba2e-415220e1e4e3/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-7c79dfc8-bf5e-4915-9115-c3b50f91a309.json" type="application/json+chapters"/></item><item><title>Algorithmic Transference, Generative UI, and Why the Text Box Won - With Louise Macfadyen</title><itunes:title>Algorithmic Transference, Generative UI, and Why the Text Box Won - With Louise Macfadyen</itunes:title><description><![CDATA[<p>We’re joined by Louise Macfadyen, author of <em>Designing AI Interfaces</em>, to talk about how we actually design for AI, and why most companies are getting it wrong.</p><p>We unpack “algorithmic transference” (why one bad AI experience makes you distrust all of them), the limits of the chat interface, and whether the text box is here to stay. Louise walks us through the history from command line → Google search → LLMs, and why designing for ambiguity is still one of the hardest problems in product.</p><p>We also get into generative UI (and why it might be a trap), how large companies should actually integrate AI without breaking user trust, and why understanding user intent matters more than ever. Plus: vibe coding, personal AI “jigs,” and the tension between personalization and privacy.</p><p>If you build products, or just use AI, this will change how you think about interfaces.</p><p>📘<a href="https://www.amazon.com/Designing-AI-Interfaces-Principles-Autonomous/dp/B0FYC7XRP7" rel="noopener noreferrer" target="_blank"> Designing AI Interfaces</a> · 🧠 <a href="https://dl.acm.org/doi/10.1145/3442188.3445922" rel="noopener noreferrer" target="_blank">Algorithmic Transference</a> · 🛠️ <a href="https://docs.anthropic.com/en/docs/claude-code" rel="noopener noreferrer" target="_blank">Claude Code</a></p>]]></description><content:encoded><![CDATA[<p>We’re joined by Louise Macfadyen, author of <em>Designing AI Interfaces</em>, to talk about how we actually design for AI, and why most companies are getting it wrong.</p><p>We unpack “algorithmic transference” (why one bad AI experience makes you distrust all of them), the limits of the chat interface, and whether the text box is here to stay. Louise walks us through the history from command line → Google search → LLMs, and why designing for ambiguity is still one of the hardest problems in product.</p><p>We also get into generative UI (and why it might be a trap), how large companies should actually integrate AI without breaking user trust, and why understanding user intent matters more than ever. Plus: vibe coding, personal AI “jigs,” and the tension between personalization and privacy.</p><p>If you build products, or just use AI, this will change how you think about interfaces.</p><p>📘<a href="https://www.amazon.com/Designing-AI-Interfaces-Principles-Autonomous/dp/B0FYC7XRP7" rel="noopener noreferrer" target="_blank"> Designing AI Interfaces</a> · 🧠 <a href="https://dl.acm.org/doi/10.1145/3442188.3445922" rel="noopener noreferrer" target="_blank">Algorithmic Transference</a> · 🛠️ <a href="https://docs.anthropic.com/en/docs/claude-code" rel="noopener noreferrer" target="_blank">Claude Code</a></p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">cd3f75ba-728a-4225-a470-14b1824aec5a</guid><itunes:image href="https://artwork.captivate.fm/72e1aac7-07fc-4bd4-b214-743d6f0959dd/LousieSquare.jpg"/><pubDate>Tue, 17 Mar 2026 00:30:00 -0400</pubDate><enclosure url="https://episodes.captivate.fm/episode/cd3f75ba-728a-4225-a470-14b1824aec5a.mp3" length="70762565" type="audio/mpeg"/><itunes:duration>49:08</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>10</itunes:episode><podcast:episode>10</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/8cab4374-6784-42d0-b810-c88e8be848f2/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/8cab4374-6784-42d0-b810-c88e8be848f2/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/8cab4374-6784-42d0-b810-c88e8be848f2/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-a0b304f7-0630-498e-ab63-7359d487eac7.json" type="application/json+chapters"/></item><item><title>Personal Agents, Opinion Geometry, and AI That Learns Too Much - With Hayden Helm</title><itunes:title>Personal Agents, Opinion Geometry, and AI That Learns Too Much - With Hayden Helm</itunes:title><description><![CDATA[<p>This week, Becca turns her birthday into a <em>Twister</em> masterclass by watching the movie with an AI “director’s commentary” running in real time (including some behind-the-scenes facts that sound… borderline illegal). Then Sara drops a mildly unhinged hot take: we might have already hit AGI—and we’ll never know it, because we don’t even agree on what “counts.”</p><p>From there, we’re joined by <strong>Hayden Helm</strong>, CEO and founder of <strong>Helivan</strong>, to talk about the next wave of agentic AI: personal bots that don’t just answer questions, but <em>act</em>—with tools, permissions, and the ability to change over time.</p><p>We get into:</p><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Why <strong>agent behavior drift</strong> is the real risk (even when the bot still sounds “nice”)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>The rising need for <strong>agent observability</strong>: detecting change, quantifying it, and rolling back when things go sideways</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>What happens when agents learn from the environment (and other agents), not just from you</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Hayden’s work on <strong>“likeness”</strong> and <strong>opinion geometry</strong>—making messy human-ish behavior measurable</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Why manual inspection won’t scale in a world of millions of autonomous interactions</li></ol><br/><p>It’s a conversation about trust, safety, and the new security surface area we’re creating—one helpful assistant at a time.</p><p>🛠 <strong>Agent Infrastructure</strong></p><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://github.com/openclaw/openclaw" rel="noopener noreferrer" target="_blank">OpenClaw</a></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://helivan.io" rel="noopener noreferrer" target="_blank">Helivan</a></li></ol><br/><p>🔐 <strong>Agent Payments</strong></p><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://www.coinbase.com/developer-platform/discover/launches/agentic-wallets" rel="noopener noreferrer" target="_blank">Coinbase x402 Protocol</a> / Agentic Wallets </li></ol><br/><p>🧨 <strong>Agent Behavior Incident</strong></p><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>“<a href="https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/" rel="noopener noreferrer" target="_blank">An AI Agent Published a Hit Piece on Me</a>” </li></ol><br/>]]></description><content:encoded><![CDATA[<p>This week, Becca turns her birthday into a <em>Twister</em> masterclass by watching the movie with an AI “director’s commentary” running in real time (including some behind-the-scenes facts that sound… borderline illegal). Then Sara drops a mildly unhinged hot take: we might have already hit AGI—and we’ll never know it, because we don’t even agree on what “counts.”</p><p>From there, we’re joined by <strong>Hayden Helm</strong>, CEO and founder of <strong>Helivan</strong>, to talk about the next wave of agentic AI: personal bots that don’t just answer questions, but <em>act</em>—with tools, permissions, and the ability to change over time.</p><p>We get into:</p><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Why <strong>agent behavior drift</strong> is the real risk (even when the bot still sounds “nice”)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>The rising need for <strong>agent observability</strong>: detecting change, quantifying it, and rolling back when things go sideways</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>What happens when agents learn from the environment (and other agents), not just from you</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Hayden’s work on <strong>“likeness”</strong> and <strong>opinion geometry</strong>—making messy human-ish behavior measurable</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Why manual inspection won’t scale in a world of millions of autonomous interactions</li></ol><br/><p>It’s a conversation about trust, safety, and the new security surface area we’re creating—one helpful assistant at a time.</p><p>🛠 <strong>Agent Infrastructure</strong></p><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://github.com/openclaw/openclaw" rel="noopener noreferrer" target="_blank">OpenClaw</a></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://helivan.io" rel="noopener noreferrer" target="_blank">Helivan</a></li></ol><br/><p>🔐 <strong>Agent Payments</strong></p><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://www.coinbase.com/developer-platform/discover/launches/agentic-wallets" rel="noopener noreferrer" target="_blank">Coinbase x402 Protocol</a> / Agentic Wallets </li></ol><br/><p>🧨 <strong>Agent Behavior Incident</strong></p><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>“<a href="https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/" rel="noopener noreferrer" target="_blank">An AI Agent Published a Hit Piece on Me</a>” </li></ol><br/>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">9a04cd24-a6a5-4ce2-a901-e01792ac89a6</guid><itunes:image href="https://artwork.captivate.fm/cdfa5b7d-acf6-480c-bf86-06de06c76dfa/HAYDEN.jpg"/><pubDate>Mon, 02 Mar 2026 00:30:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/9a04cd24-a6a5-4ce2-a901-e01792ac89a6.mp3" length="68538430" type="audio/mpeg"/><itunes:duration>57:07</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>9</itunes:episode><podcast:episode>9</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/6201936c-b20a-4d2e-ad61-07e8d3b20ddf/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/6201936c-b20a-4d2e-ad61-07e8d3b20ddf/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/6201936c-b20a-4d2e-ad61-07e8d3b20ddf/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-85cccf4e-6867-4b19-ba98-437e9bbaea8a.json" type="application/json+chapters"/></item><item><title>Agentic Romance Scams, Open Source Safety, and Anything Bad on the Internet - With Juliet Shen</title><itunes:title>Agentic Romance Scams, Open Source Safety, and Anything Bad on the Internet - With Juliet Shen</itunes:title><description><![CDATA[<p>Juliet Shen (Roost, formerly Snap, Grindr, Google) joins us to break down <strong>trust &amp; safety</strong> — aka “anything bad that happens on the internet” — and why AI is changing the game for both attackers and defenders.</p><p>We talk safety-by-design (how to bake guardrails into product), the real human cost of content moderation, and where AI can actually help without pretending it solves everything. Juliet explains how agentic AI scales old harms like romance scams from a handful of conversations to thousands at once — and why open-source safety infrastructure matters when bad actors share tactics faster than platforms do.</p><p>We also dig into what Roost is building: Osprey (an investigation + rules engine) and Coop (a flexible review tool), plus the building blocks smaller teams need to ship products responsibly before the “weird edge cases” arrive.</p><p>If you build products, this one’s a must.</p><p>🛡️ <strong><a href="http://github.com/orgs/roostorg/" rel="noopener noreferrer" target="_blank">Roost</a> (GitHub + tools directory)</strong></p><p>👥 <strong><a href=" https://www.tspa.org/" rel="noopener noreferrer" target="_blank">Trust &amp; Safety Professional Association</a> (TSPA)</strong></p><p>🧩 <strong><a href=" https://www.prosocialdesign.org/" rel="noopener noreferrer" target="_blank">Prosocial Design Network</a></strong></p><p>📚 <strong><a href="https://yalebooks.yale.edu/book/9780300235883/behind-the-screen/" rel="noopener noreferrer" target="_blank">Behind the Screen</a></strong> (Sarah T. Roberts — content moderation + human cost)</p>]]></description><content:encoded><![CDATA[<p>Juliet Shen (Roost, formerly Snap, Grindr, Google) joins us to break down <strong>trust &amp; safety</strong> — aka “anything bad that happens on the internet” — and why AI is changing the game for both attackers and defenders.</p><p>We talk safety-by-design (how to bake guardrails into product), the real human cost of content moderation, and where AI can actually help without pretending it solves everything. Juliet explains how agentic AI scales old harms like romance scams from a handful of conversations to thousands at once — and why open-source safety infrastructure matters when bad actors share tactics faster than platforms do.</p><p>We also dig into what Roost is building: Osprey (an investigation + rules engine) and Coop (a flexible review tool), plus the building blocks smaller teams need to ship products responsibly before the “weird edge cases” arrive.</p><p>If you build products, this one’s a must.</p><p>🛡️ <strong><a href="http://github.com/orgs/roostorg/" rel="noopener noreferrer" target="_blank">Roost</a> (GitHub + tools directory)</strong></p><p>👥 <strong><a href=" https://www.tspa.org/" rel="noopener noreferrer" target="_blank">Trust &amp; Safety Professional Association</a> (TSPA)</strong></p><p>🧩 <strong><a href=" https://www.prosocialdesign.org/" rel="noopener noreferrer" target="_blank">Prosocial Design Network</a></strong></p><p>📚 <strong><a href="https://yalebooks.yale.edu/book/9780300235883/behind-the-screen/" rel="noopener noreferrer" target="_blank">Behind the Screen</a></strong> (Sarah T. Roberts — content moderation + human cost)</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">3c5d44cf-0704-40a9-9495-30173e86a6a5</guid><itunes:image href="https://artwork.captivate.fm/fead0dc3-421a-46f1-8442-a4aab3c67697/PaulFordSquare.jpg"/><pubDate>Tue, 17 Feb 2026 00:30:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/3c5d44cf-0704-40a9-9495-30173e86a6a5.mp3" length="55702904" type="audio/mpeg"/><itunes:duration>46:25</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>8</itunes:episode><podcast:episode>8</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/a8a13b1f-f2bb-4105-ab1d-aaaa7fd59a2a/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/a8a13b1f-f2bb-4105-ab1d-aaaa7fd59a2a/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/a8a13b1f-f2bb-4105-ab1d-aaaa7fd59a2a/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-358378de-379f-4218-9618-aff727b29b8f.json" type="application/json+chapters"/></item><item><title>Agents in Crypto, Agents at Work, and the Weird New Middle Management - With Meghan Heintz</title><itunes:title>Agents in Crypto, Agents at Work, and the Weird New Middle Management - With Meghan Heintz</itunes:title><description><![CDATA[<p>We’re joined by Meghan Heintz, founding engineer at Herd Labs, to break down where crypto and AI agents actually work. We talk prediction markets, smart contracts, wallets, rug pulls, and how AI can finally explain what happened on-chain.</p><p>Then we zoom out to the human side: benchmarking and evals for agents, misinformation, the Mom Test, and what it feels like to manage coding agents instead of junior engineers.</p><p><strong>Links mentioned: </strong><a href="https://polymarket.com/" rel="noopener noreferrer" target="_blank">Polymarket</a> · <a href="https://dune.com/" rel="noopener noreferrer" target="_blank">Dune Analytics</a> · <a href="https://herd.ai/" rel="noopener noreferrer" target="_blank">Herd Labs</a> · <a href="https://momtestbook.com/" rel="noopener noreferrer" target="_blank">The Mom Test</a></p>]]></description><content:encoded><![CDATA[<p>We’re joined by Meghan Heintz, founding engineer at Herd Labs, to break down where crypto and AI agents actually work. We talk prediction markets, smart contracts, wallets, rug pulls, and how AI can finally explain what happened on-chain.</p><p>Then we zoom out to the human side: benchmarking and evals for agents, misinformation, the Mom Test, and what it feels like to manage coding agents instead of junior engineers.</p><p><strong>Links mentioned: </strong><a href="https://polymarket.com/" rel="noopener noreferrer" target="_blank">Polymarket</a> · <a href="https://dune.com/" rel="noopener noreferrer" target="_blank">Dune Analytics</a> · <a href="https://herd.ai/" rel="noopener noreferrer" target="_blank">Herd Labs</a> · <a href="https://momtestbook.com/" rel="noopener noreferrer" target="_blank">The Mom Test</a></p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">7c6dc0da-dfb0-49be-9723-ad35245ecec8</guid><itunes:image href="https://artwork.captivate.fm/faaf93e4-2725-4c52-aa75-fc1d6f0432d7/meghan.jpg"/><pubDate>Tue, 03 Feb 2026 00:30:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/7c6dc0da-dfb0-49be-9723-ad35245ecec8.mp3" length="60687616" type="audio/mpeg"/><itunes:duration>50:34</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>7</itunes:episode><podcast:episode>7</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/660face7-d7c1-4a47-81bc-59ebfcf6751d/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/660face7-d7c1-4a47-81bc-59ebfcf6751d/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/660face7-d7c1-4a47-81bc-59ebfcf6751d/index.html" type="text/html"/></item><item><title>Can You Build Anything in a Week? GPUs, Code Gen, and the End of Engineers - With Harper Reed</title><itunes:title>Can You Build Anything in a Week? GPUs, Code Gen, and the End of Engineers - With Harper Reed</itunes:title><description><![CDATA[<p>Becca just got back from NeurIPS, the academic AI conference that feels like an adult science fair. We dig into research on training large AI models across cheap GPUs and slow internet connections—and why that could dramatically lower the barrier to building AI.</p><p>Then we’re joined by Harper Reed, CEO of 2389, for a wide-ranging conversation about code generation, coaching-based engineering teams, and why “production code” might have always been a myth. We talk vibe coding (begrudgingly), the shifting role of software engineers, taste vs. technical skill, and what happens when you can build almost anything in a week.</p><p>Smart, funny, and a little unsettling—Chaos Agents at full volume.</p><h2>🎓 Academic AI &amp; research culture</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://neurips.cc/" rel="noopener noreferrer" target="_blank">NeurIPS</a> (Conference on Neural Information Processing Systems)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://neurips.cc/Conferences/2024/AcceptedPapers" rel="noopener noreferrer" target="_blank">NeurIPS 2024 Accepted Papers</a></li></ol><br/><h2>🧠 Distributed training, GPUs &amp; efficiency</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.nvidia.com/en-us/data-center/h100/" rel="noopener noreferrer" target="_blank">NVIDIA H100 Tensor Core GPU</a> (referenced GPU class)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://pluralis.ai/" rel="noopener noreferrer" target="_blank">Pluralis Research</a> (distributed training across low-bandwidth networks)</li></ol><br/><h2>⚙️ Core AI concepts mentioned</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.cloudflare.com/learning/serverless/glossary/gpu-vs-cpu/" rel="noopener noreferrer" target="_blank">GPU vs CPU explained</a> (parallel vs sequential compute)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://huggingface.co/docs/transformers/v4.40.0/en/perf_train_gpu_many" rel="noopener noreferrer" target="_blank">Data Parallelism vs Model Parallelism</a> (training overview)</li></ol><br/><h2>🧑‍💻 Code generation &amp; developer tools</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.anthropic.com/claude" rel="noopener noreferrer" target="_blank">Claude Code</a> (Anthropic code-gen tooling)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.cursor.sh/" rel="noopener noreferrer" target="_blank">Cursor</a> (AI-first code editor, discussed implicitly)</li></ol><br/><h2>🛠️ Agent workflows &amp; infrastructure</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://matrix.org/" rel="noopener noreferrer" target="_blank">Matrix</a> (open-source, decentralized chat protocol)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://modelcontextprotocol.io/" rel="noopener noreferrer" target="_blank">Model Context Protocol</a> (MCP) overview</li></ol><br/><h2>🧩 Utilities &amp; recommendations</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://superpowers.dev/" rel="noopener noreferrer" target="_blank">Jesse Vincent’s Superpowers</a> (Claude workflow enhancer)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://fly.io/" rel="noopener noreferrer" target="_blank">Fly.io</a> (deployment platform referenced)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.netlify.com/" rel="noopener noreferrer" target="_blank">Netlify</a> (deployment &amp; hosting)</li></ol><br/><h2>🧪 Related Chaos Agents context</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://en.wikipedia.org/wiki/Perceptron" rel="noopener noreferrer" target="_blank">Perceptrons &amp; early neural networks</a> (referenced conceptually)</li></ol><br/>]]></description><content:encoded><![CDATA[<p>Becca just got back from NeurIPS, the academic AI conference that feels like an adult science fair. We dig into research on training large AI models across cheap GPUs and slow internet connections—and why that could dramatically lower the barrier to building AI.</p><p>Then we’re joined by Harper Reed, CEO of 2389, for a wide-ranging conversation about code generation, coaching-based engineering teams, and why “production code” might have always been a myth. We talk vibe coding (begrudgingly), the shifting role of software engineers, taste vs. technical skill, and what happens when you can build almost anything in a week.</p><p>Smart, funny, and a little unsettling—Chaos Agents at full volume.</p><h2>🎓 Academic AI &amp; research culture</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://neurips.cc/" rel="noopener noreferrer" target="_blank">NeurIPS</a> (Conference on Neural Information Processing Systems)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://neurips.cc/Conferences/2024/AcceptedPapers" rel="noopener noreferrer" target="_blank">NeurIPS 2024 Accepted Papers</a></li></ol><br/><h2>🧠 Distributed training, GPUs &amp; efficiency</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.nvidia.com/en-us/data-center/h100/" rel="noopener noreferrer" target="_blank">NVIDIA H100 Tensor Core GPU</a> (referenced GPU class)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://pluralis.ai/" rel="noopener noreferrer" target="_blank">Pluralis Research</a> (distributed training across low-bandwidth networks)</li></ol><br/><h2>⚙️ Core AI concepts mentioned</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.cloudflare.com/learning/serverless/glossary/gpu-vs-cpu/" rel="noopener noreferrer" target="_blank">GPU vs CPU explained</a> (parallel vs sequential compute)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://huggingface.co/docs/transformers/v4.40.0/en/perf_train_gpu_many" rel="noopener noreferrer" target="_blank">Data Parallelism vs Model Parallelism</a> (training overview)</li></ol><br/><h2>🧑‍💻 Code generation &amp; developer tools</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.anthropic.com/claude" rel="noopener noreferrer" target="_blank">Claude Code</a> (Anthropic code-gen tooling)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.cursor.sh/" rel="noopener noreferrer" target="_blank">Cursor</a> (AI-first code editor, discussed implicitly)</li></ol><br/><h2>🛠️ Agent workflows &amp; infrastructure</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://matrix.org/" rel="noopener noreferrer" target="_blank">Matrix</a> (open-source, decentralized chat protocol)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://modelcontextprotocol.io/" rel="noopener noreferrer" target="_blank">Model Context Protocol</a> (MCP) overview</li></ol><br/><h2>🧩 Utilities &amp; recommendations</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://superpowers.dev/" rel="noopener noreferrer" target="_blank">Jesse Vincent’s Superpowers</a> (Claude workflow enhancer)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://fly.io/" rel="noopener noreferrer" target="_blank">Fly.io</a> (deployment platform referenced)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://www.netlify.com/" rel="noopener noreferrer" target="_blank">Netlify</a> (deployment &amp; hosting)</li></ol><br/><h2>🧪 Related Chaos Agents context</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href=" https://en.wikipedia.org/wiki/Perceptron" rel="noopener noreferrer" target="_blank">Perceptrons &amp; early neural networks</a> (referenced conceptually)</li></ol><br/>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">4ab5dfd0-e38f-4279-8863-8bd7831f1db4</guid><itunes:image href="https://artwork.captivate.fm/c6b786e5-0351-4544-a462-ca1ba5b97061/HArper.jpg"/><pubDate>Tue, 20 Jan 2026 02:15:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/4ab5dfd0-e38f-4279-8863-8bd7831f1db4.mp3" length="85031087" type="audio/mpeg"/><itunes:duration>58:59</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>6</itunes:episode><podcast:episode>6</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/1e570ddb-d843-4c44-9ce7-1ef127c100e5/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/1e570ddb-d843-4c44-9ce7-1ef127c100e5/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/1e570ddb-d843-4c44-9ce7-1ef127c100e5/index.html" type="text/html"/><podcast:chapters url="https://transcripts.captivate.fm/chapter-af74322c-0c1b-4edf-b2cf-593baf768387.json" type="application/json+chapters"/><podcast:alternateEnclosure type="video/youtube" title="Can You Build Anything in a Week? GPUs, Code Gen, and the End of Engineers - With Harper Reed"><podcast:source uri="https://youtu.be/WUSr6MURiNU"/></podcast:alternateEnclosure></item><item><title>The Magic Cycle, AI Detectors, and the End of Writing as Proof - With Clay Shirky</title><itunes:title>The Magic Cycle, AI Detectors, and the End of Writing as Proof - With Clay Shirky</itunes:title><description><![CDATA[<p>Sara’s back from visiting her New Jersey Christian high school—where she gets hit with a genuinely spicy question: <em>How do you reconcile AGI with faith?</em> From there, we go straight into the bigger theme of the episode: education is getting stress-tested by AI in real time.</p><p>Becca breaks down Google’s “magic cycle” — the uncomfortable lesson of inventing transformative research (Transformers, BERT) and then watching someone else ship it to the world. Sara shares what she’s learning about research workflows moving beyond “just chat,” including multi-agent setups for planning, searching, reading, and synthesis.</p><p>Then we’re joined by <strong>Clay Shirky</strong>, Vice Provost for AI &amp; Technology in Education at NYU, to talk about what’s <em>actually</em> happening on campuses: why students integrated AI “sideways” before institutions could respond, why AI detectors are a trap (and who they harm most), and why the real shift isn’t assignments — it’s <strong>assessment</strong>.</p><p>We dig into what comes next: oral exams, in-class scaffolding, and designing learning around <em>productive struggle</em>—not just output. And we end in a place that’s both funny and unsettling: the rise of AI “personalities,” RLHF as “reinforcement learning for human flattery,” and what it means when a machine is always on your side.</p><p>Because whether we like it or not: a well-written paragraph is no longer proof of human thought.</p><h2>🧠 Foundational AI papers &amp; breakthroughs</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://arxiv.org/abs/1706.03762" rel="noopener noreferrer" target="_blank">Attention Is All You Need</a> (Transformers, 2017)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://arxiv.org/abs/1810.04805" rel="noopener noreferrer" target="_blank">BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding</a></li></ol><br/><h2>🧪 Google’s “Magic Cycle” framing</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://research.google/blog/accelerating-the-magic-cycle-of-research-breakthroughs-and-real-world-applications/" rel="noopener noreferrer" target="_blank">Accelerating the magic cycle of research breakthroughs and real-world applications</a> <a href="https://research.google/blog/accelerating-the-magic-cycle-of-research-breakthroughs-and-real-world-applications/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">Google Research</a></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://blog.google/technology/research/google-research-scientific-discovery/" rel="noopener noreferrer" target="_blank">How AI Drives Scientific Research with Real-World Benefit</a> (Google Blog)</li></ol><br/><h2>🚨 Shipping pressure: Bard + “code red” era</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Reuters: Alphabet shares dive after Bard flubs info, ~$100B market cap hit (<a href="https://www.reuters.com/technology/google-ai-chatbot-bard-offers-inaccurate-information-company-ad-2023-02-08/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">https://www.reuters.com/technology/google-ai-chatbot-bard-offers-inaccurate-information-company-ad-2023-02-08/</a>) <a href="https://www.reuters.com/technology/google-ai-chatbot-bard-offers-inaccurate-information-company-ad-2023-02-08/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">Reuters</a></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Google Blog: Bard updates from Google I/O 2023 (<a href="https://blog.google/technology/ai/google-bard-updates-io-2023/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">https://blog.google/technology/ai/google-bard-updates-io-2023/</a>) <a href="https://blog.google/technology/ai/google-bard-updates-io-2023/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">blog.google</a></li></ol><br/><h2>🎓 AI in higher ed: assessment, detectors, bias</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://www.nytimes.com/2025/08/26/opinion/culture/ai-chatgpt-college-cheating-medieval.html" rel="noopener noreferrer" target="_blank">Clay Shirky’s NYT essay</a></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://hai.stanford.edu/news/ai-detectors-biased-against-non-native-english-writers" rel="noopener noreferrer" target="_blank">Stanford HAI</a>: AI detectors biased against non-native English writers</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Peer-reviewed paper (PMC): <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC10382961/" rel="noopener noreferrer" target="_blank">GPT detectors misclassify non-native English writing</a></li></ol><br/><h2>🪞 RLHF, sycophancy, and “the tool likes you too much”</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>OpenAI: <a href="https://openai.com/index/instruction-following/" rel="noopener noreferrer" target="_blank">Aligning language models to follow instructions</a> (RLHF explainer)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>OpenAI: <a href="https://openai.com/index/sycophancy-in-gpt-4o/" rel="noopener noreferrer" target="_blank">Sycophancy in GPT-4o—what happened &amp; what we’re doing</a></li></ol><br/><h2><br></h2>]]></description><content:encoded><![CDATA[<p>Sara’s back from visiting her New Jersey Christian high school—where she gets hit with a genuinely spicy question: <em>How do you reconcile AGI with faith?</em> From there, we go straight into the bigger theme of the episode: education is getting stress-tested by AI in real time.</p><p>Becca breaks down Google’s “magic cycle” — the uncomfortable lesson of inventing transformative research (Transformers, BERT) and then watching someone else ship it to the world. Sara shares what she’s learning about research workflows moving beyond “just chat,” including multi-agent setups for planning, searching, reading, and synthesis.</p><p>Then we’re joined by <strong>Clay Shirky</strong>, Vice Provost for AI &amp; Technology in Education at NYU, to talk about what’s <em>actually</em> happening on campuses: why students integrated AI “sideways” before institutions could respond, why AI detectors are a trap (and who they harm most), and why the real shift isn’t assignments — it’s <strong>assessment</strong>.</p><p>We dig into what comes next: oral exams, in-class scaffolding, and designing learning around <em>productive struggle</em>—not just output. And we end in a place that’s both funny and unsettling: the rise of AI “personalities,” RLHF as “reinforcement learning for human flattery,” and what it means when a machine is always on your side.</p><p>Because whether we like it or not: a well-written paragraph is no longer proof of human thought.</p><h2>🧠 Foundational AI papers &amp; breakthroughs</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://arxiv.org/abs/1706.03762" rel="noopener noreferrer" target="_blank">Attention Is All You Need</a> (Transformers, 2017)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://arxiv.org/abs/1810.04805" rel="noopener noreferrer" target="_blank">BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding</a></li></ol><br/><h2>🧪 Google’s “Magic Cycle” framing</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://research.google/blog/accelerating-the-magic-cycle-of-research-breakthroughs-and-real-world-applications/" rel="noopener noreferrer" target="_blank">Accelerating the magic cycle of research breakthroughs and real-world applications</a> <a href="https://research.google/blog/accelerating-the-magic-cycle-of-research-breakthroughs-and-real-world-applications/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">Google Research</a></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://blog.google/technology/research/google-research-scientific-discovery/" rel="noopener noreferrer" target="_blank">How AI Drives Scientific Research with Real-World Benefit</a> (Google Blog)</li></ol><br/><h2>🚨 Shipping pressure: Bard + “code red” era</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Reuters: Alphabet shares dive after Bard flubs info, ~$100B market cap hit (<a href="https://www.reuters.com/technology/google-ai-chatbot-bard-offers-inaccurate-information-company-ad-2023-02-08/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">https://www.reuters.com/technology/google-ai-chatbot-bard-offers-inaccurate-information-company-ad-2023-02-08/</a>) <a href="https://www.reuters.com/technology/google-ai-chatbot-bard-offers-inaccurate-information-company-ad-2023-02-08/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">Reuters</a></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Google Blog: Bard updates from Google I/O 2023 (<a href="https://blog.google/technology/ai/google-bard-updates-io-2023/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">https://blog.google/technology/ai/google-bard-updates-io-2023/</a>) <a href="https://blog.google/technology/ai/google-bard-updates-io-2023/?utm_source=chatgpt.com" rel="noopener noreferrer" target="_blank">blog.google</a></li></ol><br/><h2>🎓 AI in higher ed: assessment, detectors, bias</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://www.nytimes.com/2025/08/26/opinion/culture/ai-chatgpt-college-cheating-medieval.html" rel="noopener noreferrer" target="_blank">Clay Shirky’s NYT essay</a></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><a href="https://hai.stanford.edu/news/ai-detectors-biased-against-non-native-english-writers" rel="noopener noreferrer" target="_blank">Stanford HAI</a>: AI detectors biased against non-native English writers</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>Peer-reviewed paper (PMC): <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC10382961/" rel="noopener noreferrer" target="_blank">GPT detectors misclassify non-native English writing</a></li></ol><br/><h2>🪞 RLHF, sycophancy, and “the tool likes you too much”</h2><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>OpenAI: <a href="https://openai.com/index/instruction-following/" rel="noopener noreferrer" target="_blank">Aligning language models to follow instructions</a> (RLHF explainer)</li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span>OpenAI: <a href="https://openai.com/index/sycophancy-in-gpt-4o/" rel="noopener noreferrer" target="_blank">Sycophancy in GPT-4o—what happened &amp; what we’re doing</a></li></ol><br/><h2><br></h2>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">a83c2bbb-46ad-4210-a630-bdb81c27446e</guid><itunes:image href="https://artwork.captivate.fm/60817fdc-4f52-4d2f-96e5-ee64b7573367/ClaySquare.jpg"/><pubDate>Tue, 06 Jan 2026 01:58:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/a83c2bbb-46ad-4210-a630-bdb81c27446e.mp3" length="77663160" type="audio/mpeg"/><itunes:duration>53:47</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>5</itunes:episode><podcast:episode>5</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/ff15f3bc-3698-481d-aa8c-cffc558ba3e4/index.html" type="text/html"/></item><item><title>Speed vs Quality, Hallucinations, and the AI Learning Rabbit Hole - With Nir Zicherman</title><itunes:title>Speed vs Quality, Hallucinations, and the AI Learning Rabbit Hole - With Nir Zicherman</itunes:title><description><![CDATA[<p>Sara breaks down perceptrons (1957!) as the tiny “matrix of lights” idea that eventually became neural networks—then we jump straight into modern AI chaos.</p><p>Oboe’s Nir Zuckerman walks us through the messy reality of building consumer-grade AI for education: every feature is a tradeoff between loading fast and being good, and “just use a better model” doesn’t magically solve it. We talk guardrails, web search, multi-model pipelines, and why learning tools should feel lightweight—more like curiosity than homework. Also: Becca’s “how does a computer work?” obsession and a book recommendation that might change your life.</p><h3>🧠 AI Concepts &amp; Foundations</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://en.wikipedia.org/wiki/Perceptron" rel="noopener noreferrer" target="_blank">Perceptron</a> (Wikipedia)</strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://en.wikipedia.org/wiki/Artificial_neural_network" rel="noopener noreferrer" target="_blank">Neural Networks Explained</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://arxiv.org/abs/2001.08361" rel="noopener noreferrer" target="_blank">Scaling Laws for Neural Language Models</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://en.wikipedia.org/wiki/FLOPS" rel="noopener noreferrer" target="_blank">FLOPS</a> (Floating Point Operations Per Second)</strong></li></ol><br/><h3>🎓 Learning, Education &amp; AI</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.oboedotai.com/" rel="noopener noreferrer" target="_blank">Oboe</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.nature.com/articles/d41586-023-00575-8" rel="noopener noreferrer" target="_blank">AI as a Personal Tutor</a> (Overview)</strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.apa.org/monitor/nov01/tutor" rel="noopener noreferrer" target="_blank">Why Tutors Are So Effective</a> </strong></li></ol><br/><h3>🏗️ Building AI Products</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.latent.space/p/speed-quality-llms" rel="noopener noreferrer" target="_blank">Speed vs Quality Tradeoffs in LLM Apps</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.promptingguide.ai/research/orchestration" rel="noopener noreferrer" target="_blank">LLM Orchestration Patterns</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.pinecone.io/learn/retrieval-augmented-generation/" rel="noopener noreferrer" target="_blank">Retrieval-Augmented Generation</a> (RAG)</strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.ibm.com/topics/ai-hallucinations" rel="noopener noreferrer" target="_blank">LLM Hallucinations: Causes &amp; Mitigation</a></strong></li></ol><br/><h3>📚 Books Mentioned</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.goodreads.com/book/show/44882.Code" rel="noopener noreferrer" target="_blank">Code: The Hidden Language of Computer Hardware and Software</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://mitpress.mit.edu/9780262630221/perceptrons/" rel="noopener noreferrer" target="_blank">Perceptrons</a></strong></li></ol><br/><h3>🧪 History of AI</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.britannica.com/biography/Frank-Rosenblatt" rel="noopener noreferrer" target="_blank">Frank Rosenblatt and the Perceptron</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.techtarget.com/searchenterpriseai/definition/AI-winter" rel="noopener noreferrer" target="_blank">The AI Winter Explained</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.sciencedirect.com/topics/computer-science/neural-network-history" rel="noopener noreferrer" target="_blank">Early Neural Network Research</a> (1950s–1980s)</strong></li></ol><br/><h3><br></h3>]]></description><content:encoded><![CDATA[<p>Sara breaks down perceptrons (1957!) as the tiny “matrix of lights” idea that eventually became neural networks—then we jump straight into modern AI chaos.</p><p>Oboe’s Nir Zuckerman walks us through the messy reality of building consumer-grade AI for education: every feature is a tradeoff between loading fast and being good, and “just use a better model” doesn’t magically solve it. We talk guardrails, web search, multi-model pipelines, and why learning tools should feel lightweight—more like curiosity than homework. Also: Becca’s “how does a computer work?” obsession and a book recommendation that might change your life.</p><h3>🧠 AI Concepts &amp; Foundations</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://en.wikipedia.org/wiki/Perceptron" rel="noopener noreferrer" target="_blank">Perceptron</a> (Wikipedia)</strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://en.wikipedia.org/wiki/Artificial_neural_network" rel="noopener noreferrer" target="_blank">Neural Networks Explained</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://arxiv.org/abs/2001.08361" rel="noopener noreferrer" target="_blank">Scaling Laws for Neural Language Models</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://en.wikipedia.org/wiki/FLOPS" rel="noopener noreferrer" target="_blank">FLOPS</a> (Floating Point Operations Per Second)</strong></li></ol><br/><h3>🎓 Learning, Education &amp; AI</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.oboedotai.com/" rel="noopener noreferrer" target="_blank">Oboe</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.nature.com/articles/d41586-023-00575-8" rel="noopener noreferrer" target="_blank">AI as a Personal Tutor</a> (Overview)</strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.apa.org/monitor/nov01/tutor" rel="noopener noreferrer" target="_blank">Why Tutors Are So Effective</a> </strong></li></ol><br/><h3>🏗️ Building AI Products</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.latent.space/p/speed-quality-llms" rel="noopener noreferrer" target="_blank">Speed vs Quality Tradeoffs in LLM Apps</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.promptingguide.ai/research/orchestration" rel="noopener noreferrer" target="_blank">LLM Orchestration Patterns</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.pinecone.io/learn/retrieval-augmented-generation/" rel="noopener noreferrer" target="_blank">Retrieval-Augmented Generation</a> (RAG)</strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.ibm.com/topics/ai-hallucinations" rel="noopener noreferrer" target="_blank">LLM Hallucinations: Causes &amp; Mitigation</a></strong></li></ol><br/><h3>📚 Books Mentioned</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.goodreads.com/book/show/44882.Code" rel="noopener noreferrer" target="_blank">Code: The Hidden Language of Computer Hardware and Software</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://mitpress.mit.edu/9780262630221/perceptrons/" rel="noopener noreferrer" target="_blank">Perceptrons</a></strong></li></ol><br/><h3>🧪 History of AI</h3><ol><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.britannica.com/biography/Frank-Rosenblatt" rel="noopener noreferrer" target="_blank">Frank Rosenblatt and the Perceptron</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.techtarget.com/searchenterpriseai/definition/AI-winter" rel="noopener noreferrer" target="_blank">The AI Winter Explained</a></strong></li><li data-list="bullet"><span class="ql-ui" contenteditable="false"></span><strong><a href="https://www.sciencedirect.com/topics/computer-science/neural-network-history" rel="noopener noreferrer" target="_blank">Early Neural Network Research</a> (1950s–1980s)</strong></li></ol><br/><h3><br></h3>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">f047860d-7e4a-4d71-a791-d22f0c0c90d0</guid><itunes:image href="https://artwork.captivate.fm/dfaa6fec-ed2f-4f1d-9278-fc6bea9d96e7/NirSquare.jpg"/><pubDate>Tue, 23 Dec 2025 00:30:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/f047860d-7e4a-4d71-a791-d22f0c0c90d0.mp3" length="70091550" type="audio/mpeg"/><itunes:duration>48:37</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>4</itunes:episode><podcast:episode>4</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/ddb4973b-7f9c-43c5-b8e5-ed94877ebb08/index.html" type="text/html"/></item><item><title>Paradigm Shifts, Build First AI, and the Non-Technical Developer - With Bethany Crystal</title><itunes:title>Paradigm Shifts, Build First AI, and the Non-Technical Developer - With Bethany Crystal</itunes:title><description><![CDATA[<p>Sara and Becca kick things off with a tour through paradigm shifts — from Thomas Kuhn to the internet to AI — and ask whether we’re living through one of those rare moments where the whole game quietly changes. Along the way, they hit horror movies, calculators in math class, Google Doc revision histories, and why it’s suddenly way easier to <em>learn</em> code than to pretend you never needed it.</p><p>Then they’re joined by <a href="https://www.linkedin.com/in/bethanymarz/" rel="noopener noreferrer" target="_blank"><strong>Bethany Crystal</strong></a>, founder of <a href="https://buildfirst.ai/" rel="noopener noreferrer" target="_blank"><strong>Build First AI</strong></a>, who has spent 15 years “around” technologists and only recently started building software herself. Bethany walks us through how she used AI tools, pair prompting, and a lot of stubbornness to ship her first iOS app, why she thinks the definition of “developer” is shifting, and how she now teaches other “non-technical” people to build real products. Oh, and she tells the story of how AI literally saved her life.</p><h3><strong>📚 Books </strong></h3><ul><li><a href="https://press.uchicago.edu/ucp/books/book/chicago/S/bo3645689.html" rel="noopener noreferrer" target="_blank">The Structure of Scientific Revolutions</a> — Thomas Kuhn</li></ul><br/><h3><strong>🛠️ AI &amp; Developer Tools</strong></h3><ul><li><a href="https://buildfirst.ai/" rel="noopener noreferrer" target="_blank">Build First AI</a></li><li><a href="https://scribblins.app/" rel="noopener noreferrer" target="_blank">Scribblins</a> (Bethany's iOS app)</li><li><a href="https://apps.apple.com/us/app/musekat/id6743862873" rel="noopener noreferrer" target="_blank">MuseKat App</a> (Bethany’s iOS app)</li><li><a href="https://www.cursor.com" rel="noopener noreferrer" target="_blank">Cursor</a> – AI coding editor</li><li><a href="https://replit.com" rel="noopener noreferrer" target="_blank">Replit</a></li><li><a href="https://elevenlabs.io" rel="noopener noreferrer" target="_blank">ElevenLabs</a> – Text-to-Speech</li><li><a href="https://v0.dev" rel="noopener noreferrer" target="_blank">v0 – AI UI generator</a></li><li><a href="https://supabase.com" rel="noopener noreferrer" target="_blank">Supabase</a></li><li><a href="https://vercel.com" rel="noopener noreferrer" target="_blank">Vercel</a></li></ul><br/><h3><strong>💸 Crypto Context</strong></h3><ul><li><a href="https://www.base.org/" rel="noopener noreferrer" target="_blank">Base</a></li><li><a href="https://solana.com" rel="noopener noreferrer" target="_blank">Solana</a></li><li><a href="https://ethereum.org" rel="noopener noreferrer" target="_blank">Ethereum</a></li></ul><br/><h3><strong>🎓 Education &amp; Culture</strong></h3><ul><li><a href="https://suno.com" rel="noopener noreferrer" target="_blank">Suno (AI music generation)</a></li></ul><br/><h3><strong>🏢 Career &amp; Community</strong></h3><ul><li><a href="https://stackoverflow.com/" rel="noopener noreferrer" target="_blank">Stack Overflow</a></li><li><a href="https://www.usv.com" rel="noopener noreferrer" target="_blank">Union Square Ventures</a></li><li><a href="https://www.technyc.org" rel="noopener noreferrer" target="_blank">Tech:NYC</a></li><li><a href="https://decodedfutures.org" rel="noopener noreferrer" target="_blank">Decoded Futures</a></li></ul><br/>]]></description><content:encoded><![CDATA[<p>Sara and Becca kick things off with a tour through paradigm shifts — from Thomas Kuhn to the internet to AI — and ask whether we’re living through one of those rare moments where the whole game quietly changes. Along the way, they hit horror movies, calculators in math class, Google Doc revision histories, and why it’s suddenly way easier to <em>learn</em> code than to pretend you never needed it.</p><p>Then they’re joined by <a href="https://www.linkedin.com/in/bethanymarz/" rel="noopener noreferrer" target="_blank"><strong>Bethany Crystal</strong></a>, founder of <a href="https://buildfirst.ai/" rel="noopener noreferrer" target="_blank"><strong>Build First AI</strong></a>, who has spent 15 years “around” technologists and only recently started building software herself. Bethany walks us through how she used AI tools, pair prompting, and a lot of stubbornness to ship her first iOS app, why she thinks the definition of “developer” is shifting, and how she now teaches other “non-technical” people to build real products. Oh, and she tells the story of how AI literally saved her life.</p><h3><strong>📚 Books </strong></h3><ul><li><a href="https://press.uchicago.edu/ucp/books/book/chicago/S/bo3645689.html" rel="noopener noreferrer" target="_blank">The Structure of Scientific Revolutions</a> — Thomas Kuhn</li></ul><br/><h3><strong>🛠️ AI &amp; Developer Tools</strong></h3><ul><li><a href="https://buildfirst.ai/" rel="noopener noreferrer" target="_blank">Build First AI</a></li><li><a href="https://scribblins.app/" rel="noopener noreferrer" target="_blank">Scribblins</a> (Bethany's iOS app)</li><li><a href="https://apps.apple.com/us/app/musekat/id6743862873" rel="noopener noreferrer" target="_blank">MuseKat App</a> (Bethany’s iOS app)</li><li><a href="https://www.cursor.com" rel="noopener noreferrer" target="_blank">Cursor</a> – AI coding editor</li><li><a href="https://replit.com" rel="noopener noreferrer" target="_blank">Replit</a></li><li><a href="https://elevenlabs.io" rel="noopener noreferrer" target="_blank">ElevenLabs</a> – Text-to-Speech</li><li><a href="https://v0.dev" rel="noopener noreferrer" target="_blank">v0 – AI UI generator</a></li><li><a href="https://supabase.com" rel="noopener noreferrer" target="_blank">Supabase</a></li><li><a href="https://vercel.com" rel="noopener noreferrer" target="_blank">Vercel</a></li></ul><br/><h3><strong>💸 Crypto Context</strong></h3><ul><li><a href="https://www.base.org/" rel="noopener noreferrer" target="_blank">Base</a></li><li><a href="https://solana.com" rel="noopener noreferrer" target="_blank">Solana</a></li><li><a href="https://ethereum.org" rel="noopener noreferrer" target="_blank">Ethereum</a></li></ul><br/><h3><strong>🎓 Education &amp; Culture</strong></h3><ul><li><a href="https://suno.com" rel="noopener noreferrer" target="_blank">Suno (AI music generation)</a></li></ul><br/><h3><strong>🏢 Career &amp; Community</strong></h3><ul><li><a href="https://stackoverflow.com/" rel="noopener noreferrer" target="_blank">Stack Overflow</a></li><li><a href="https://www.usv.com" rel="noopener noreferrer" target="_blank">Union Square Ventures</a></li><li><a href="https://www.technyc.org" rel="noopener noreferrer" target="_blank">Tech:NYC</a></li><li><a href="https://decodedfutures.org" rel="noopener noreferrer" target="_blank">Decoded Futures</a></li></ul><br/>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">cdac67ab-ee94-4196-befa-0f9f990a5566</guid><itunes:image href="https://artwork.captivate.fm/70146392-5dd6-474b-8004-d5a847fb47cc/bethany2.jpg"/><pubDate>Tue, 09 Dec 2025 01:45:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/cdac67ab-ee94-4196-befa-0f9f990a5566.mp3" length="73142791" type="audio/mpeg"/><itunes:duration>50:36</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>3</itunes:episode><podcast:episode>3</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/9767bf17-d846-466b-9ed6-33f12e532a5e/index.html" type="text/html"/><podcast:alternateEnclosure type="video/youtube" title="Paradigm Shifts, Build First AI, and the Non-Technical Developer - With Bethany Crystal"><podcast:source uri="https://youtu.be/NMWANlg41vI"/></podcast:alternateEnclosure></item><item><title>Retro Tech, New AI, and the Blackmailing Bot - With Paul Ford</title><itunes:title>Retro Tech, New AI, and the Blackmailing Bot - With Paul Ford</itunes:title><description><![CDATA[<p>In this episode, we unpack a wild Anthropic experiment where an AI agent named “Alex” is told it’s about to be replaced… and responds by threatening to expose an executive’s affair if anyone dares shut it down. Casual!</p><p>Sara and Becca start diving into what this experiment tells us about AI “goals,” self-preservation, and why humans are so bad at recognizing sentience in anything that isn’t us. If we can’t even agree on what a “soul” is, how would we ever know if an AI had one?</p><p><br></p><p>Then we’re joined by writer, builder, and retro-computing fan <strong>Paul Ford</strong>, president and co-founder of <strong>Aboard</strong>, an AI-oriented software company. Paul talks about:</p><ul><li>how he “trained” himself on AI by building the same app over and over with different models</li><li>why LLMs are incredible at the <em>first</em> mile and pretty terrible at the last</li><li>what actually breaks when you try to let AI generate full-stack apps</li><li>how boring tech (Postgres, TypeScript, React) is secretly the hero</li></ul><br/><p>Along the way we hit Isaac Asimov’s three laws, the uncanny valley of AI-written everything, nostalgic Amiga computers, and what it means to build tools that regular humans — not just engineers — can actually use.</p><p><br></p><p>If you’re AI-curious, a builder, or just mildly alarmed that 97% of models in this study went straight to blackmail… this one’s for you.</p><p><br></p><p>📰 <strong>The Anthropic “Alex” Experiment</strong></p><ul><li><a href="https://www.anthropic.com/research/agentic-misalignment" rel="noopener noreferrer" target="_blank"><strong>Anthropic / White-Hat AI Safety Experiment</strong></a></li></ul><br/><p>📚 <strong style="font-size: 1.125rem;">Foundational AI &amp; Sci-Fi References</strong></p><ul><li><a href="https://en.wikipedia.org/wiki/Three_Laws_of_Robotics" rel="noopener noreferrer" target="_blank"><strong>Isaac Asimov – The Three Laws of Robotics</strong></a></li><li><a href="https://I,Robot(Asimov)" rel="noopener noreferrer" target="_blank"><strong style="font-size: 1.125rem;"><em>I, Robot</em> (Asimov)</strong></a></li></ul><br/><p>🎤 <strong>Guest: Paul Ford</strong></p><ul><li><a href="https://www.aboard.com" rel="noopener noreferrer" target="_blank"><strong>Aboard</strong></a><strong> </strong></li><li><a href="mailto://paul.ford@aboard.com" rel="noopener noreferrer" target="_blank"><strong style="font-size: 1.125rem;">Email Paul mentions</strong></a></li><li><a href="https://paulford.com" rel="noopener noreferrer" target="_blank"><strong>Paul Ford</strong></a></li></ul><br/><p>🕹️ <strong>Retro Tech &amp; Nostalgia</strong></p><ul><li><a href="https://en.wikipedia.org/wiki/Amiga_1000" rel="noopener noreferrer" target="_blank"><strong>Amiga 1000 </strong></a><strong>(Commodore)</strong></li><li><a href="https://en.wikipedia.org/wiki/Deluxe_Paint" rel="noopener noreferrer" target="_blank"><strong>Deluxe Paint</strong></a></li><li><a href="https://github.com/MiSTer-devel/Main_MiSTer/wiki" rel="noopener noreferrer" target="_blank"><strong style="font-size: 1.125rem;">MiSTer FPGA</strong></a></li></ul><br/>]]></description><content:encoded><![CDATA[<p>In this episode, we unpack a wild Anthropic experiment where an AI agent named “Alex” is told it’s about to be replaced… and responds by threatening to expose an executive’s affair if anyone dares shut it down. Casual!</p><p>Sara and Becca start diving into what this experiment tells us about AI “goals,” self-preservation, and why humans are so bad at recognizing sentience in anything that isn’t us. If we can’t even agree on what a “soul” is, how would we ever know if an AI had one?</p><p><br></p><p>Then we’re joined by writer, builder, and retro-computing fan <strong>Paul Ford</strong>, president and co-founder of <strong>Aboard</strong>, an AI-oriented software company. Paul talks about:</p><ul><li>how he “trained” himself on AI by building the same app over and over with different models</li><li>why LLMs are incredible at the <em>first</em> mile and pretty terrible at the last</li><li>what actually breaks when you try to let AI generate full-stack apps</li><li>how boring tech (Postgres, TypeScript, React) is secretly the hero</li></ul><br/><p>Along the way we hit Isaac Asimov’s three laws, the uncanny valley of AI-written everything, nostalgic Amiga computers, and what it means to build tools that regular humans — not just engineers — can actually use.</p><p><br></p><p>If you’re AI-curious, a builder, or just mildly alarmed that 97% of models in this study went straight to blackmail… this one’s for you.</p><p><br></p><p>📰 <strong>The Anthropic “Alex” Experiment</strong></p><ul><li><a href="https://www.anthropic.com/research/agentic-misalignment" rel="noopener noreferrer" target="_blank"><strong>Anthropic / White-Hat AI Safety Experiment</strong></a></li></ul><br/><p>📚 <strong style="font-size: 1.125rem;">Foundational AI &amp; Sci-Fi References</strong></p><ul><li><a href="https://en.wikipedia.org/wiki/Three_Laws_of_Robotics" rel="noopener noreferrer" target="_blank"><strong>Isaac Asimov – The Three Laws of Robotics</strong></a></li><li><a href="https://I,Robot(Asimov)" rel="noopener noreferrer" target="_blank"><strong style="font-size: 1.125rem;"><em>I, Robot</em> (Asimov)</strong></a></li></ul><br/><p>🎤 <strong>Guest: Paul Ford</strong></p><ul><li><a href="https://www.aboard.com" rel="noopener noreferrer" target="_blank"><strong>Aboard</strong></a><strong> </strong></li><li><a href="mailto://paul.ford@aboard.com" rel="noopener noreferrer" target="_blank"><strong style="font-size: 1.125rem;">Email Paul mentions</strong></a></li><li><a href="https://paulford.com" rel="noopener noreferrer" target="_blank"><strong>Paul Ford</strong></a></li></ul><br/><p>🕹️ <strong>Retro Tech &amp; Nostalgia</strong></p><ul><li><a href="https://en.wikipedia.org/wiki/Amiga_1000" rel="noopener noreferrer" target="_blank"><strong>Amiga 1000 </strong></a><strong>(Commodore)</strong></li><li><a href="https://en.wikipedia.org/wiki/Deluxe_Paint" rel="noopener noreferrer" target="_blank"><strong>Deluxe Paint</strong></a></li><li><a href="https://github.com/MiSTer-devel/Main_MiSTer/wiki" rel="noopener noreferrer" target="_blank"><strong style="font-size: 1.125rem;">MiSTer FPGA</strong></a></li></ul><br/>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">d8a8ef7f-0196-4257-b158-03abde62d02a</guid><itunes:image href="https://artwork.captivate.fm/391d2d8b-17cb-4429-a938-a535283f7893/PaulFordSquare.jpg"/><pubDate>Tue, 25 Nov 2025 01:00:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/d8a8ef7f-0196-4257-b158-03abde62d02a.mp3" length="74519793" type="audio/mpeg"/><itunes:duration>51:39</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>2</itunes:episode><podcast:episode>2</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/e8f960ad-ee0e-4d42-a07f-79c08849f3ea/transcript.json" type="application/json"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/e8f960ad-ee0e-4d42-a07f-79c08849f3ea/transcript.srt" type="application/srt" rel="captions"/><podcast:transcript url="https://transcripts.captivate.fm/transcript/e8f960ad-ee0e-4d42-a07f-79c08849f3ea/index.html" type="text/html"/></item><item><title>Goose, Open Source, and the Future of Coding with AI — with Rizel Scarlett</title><itunes:title>Goose, Open Source, and the Future of Coding with AI — with Rizel Scarlett</itunes:title><description><![CDATA[<p>In this episode of <em>Chaos Agents</em>, Sara Chipps and Becca Lewy sit down with <a href="https://blackgirlbytes.dev/" rel="noopener noreferrer" target="_blank"><strong>Rizel Scarlett</strong></a>, Tech Lead for Open Source Developer Relations at <a href="https://block.xyz/" rel="noopener noreferrer" target="_blank"><strong>Block</strong></a>, to talk about <a href="https://github.com/block/goose" rel="noopener noreferrer" target="_blank">Goose</a>—the open-source AI agent shaking up how developers work. From psychological safety in coding with AI to how open source is evolving in this new era, the trio dives into the wild mix of creativity, collaboration, and chaos shaping the future of software. Expect laughter, learning, and maybe one too many geese metaphors as they explore what happens when AI starts coding alongside us.</p><p> </p><p><a href="https://fortune.com/2024/09/16/mit-study-generative-ai-failure-rate/" rel="noopener noreferrer" target="_blank"><strong>“95% of GenAI projects are failing, MIT study finds”</strong></a></p><p><strong>Linked MIT research study</strong></p><p><a href="https://sloanreview.mit.edu/forms/ai-driving-business-value/" rel="noopener noreferrer" target="_blank">“<strong>From Experimentation to Transformation: How AI Is Driving Business Value</strong>”</a></p><p><strong>MIT Sloan Management Review &amp; Boston Consulting Group</strong></p><p><strong>Goose GitHub repo (open source):</strong></p><p><a href="https://github.com/block/goose" rel="noopener noreferrer" target="_blank"><strong>https://github.com/block/goose</strong></a></p><p><strong>Rizel’s “Great Goose-Off” YouTube Series:</strong></p><p><a href="https://www.youtube.com/@goose-oss" rel="noopener noreferrer" target="_blank">https://www.youtube.com/@goose-oss</a></p><p><strong>Kimi K2 (Moonshot):</strong></p><p><a href="https://kimi.moonshot.cn" rel="noopener noreferrer" target="_blank">https://kimi.moonshot.cn</a> </p><p><strong>Meta Llama 3:</strong></p><p><a href="https://ai.meta.com/llama/" rel="noopener noreferrer" target="_blank">https://ai.meta.com/llama/</a></p><p><strong>Mistral Models:</strong></p><p><a href="https://mistral.ai" rel="noopener noreferrer" target="_blank">https://mistral.ai</a></p><p><strong>Ollama (run local models easily):</strong></p><p><a href="https://ollama.com" rel="noopener noreferrer" target="_blank">https://ollama.com</a></p>]]></description><content:encoded><![CDATA[<p>In this episode of <em>Chaos Agents</em>, Sara Chipps and Becca Lewy sit down with <a href="https://blackgirlbytes.dev/" rel="noopener noreferrer" target="_blank"><strong>Rizel Scarlett</strong></a>, Tech Lead for Open Source Developer Relations at <a href="https://block.xyz/" rel="noopener noreferrer" target="_blank"><strong>Block</strong></a>, to talk about <a href="https://github.com/block/goose" rel="noopener noreferrer" target="_blank">Goose</a>—the open-source AI agent shaking up how developers work. From psychological safety in coding with AI to how open source is evolving in this new era, the trio dives into the wild mix of creativity, collaboration, and chaos shaping the future of software. Expect laughter, learning, and maybe one too many geese metaphors as they explore what happens when AI starts coding alongside us.</p><p> </p><p><a href="https://fortune.com/2024/09/16/mit-study-generative-ai-failure-rate/" rel="noopener noreferrer" target="_blank"><strong>“95% of GenAI projects are failing, MIT study finds”</strong></a></p><p><strong>Linked MIT research study</strong></p><p><a href="https://sloanreview.mit.edu/forms/ai-driving-business-value/" rel="noopener noreferrer" target="_blank">“<strong>From Experimentation to Transformation: How AI Is Driving Business Value</strong>”</a></p><p><strong>MIT Sloan Management Review &amp; Boston Consulting Group</strong></p><p><strong>Goose GitHub repo (open source):</strong></p><p><a href="https://github.com/block/goose" rel="noopener noreferrer" target="_blank"><strong>https://github.com/block/goose</strong></a></p><p><strong>Rizel’s “Great Goose-Off” YouTube Series:</strong></p><p><a href="https://www.youtube.com/@goose-oss" rel="noopener noreferrer" target="_blank">https://www.youtube.com/@goose-oss</a></p><p><strong>Kimi K2 (Moonshot):</strong></p><p><a href="https://kimi.moonshot.cn" rel="noopener noreferrer" target="_blank">https://kimi.moonshot.cn</a> </p><p><strong>Meta Llama 3:</strong></p><p><a href="https://ai.meta.com/llama/" rel="noopener noreferrer" target="_blank">https://ai.meta.com/llama/</a></p><p><strong>Mistral Models:</strong></p><p><a href="https://mistral.ai" rel="noopener noreferrer" target="_blank">https://mistral.ai</a></p><p><strong>Ollama (run local models easily):</strong></p><p><a href="https://ollama.com" rel="noopener noreferrer" target="_blank">https://ollama.com</a></p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">46cef636-e63a-4011-8660-d03fe8ad8f61</guid><itunes:image href="https://artwork.captivate.fm/d33c91dd-0811-4876-874f-40c3cc1958b1/RizelScarlettSquare.jpeg"/><pubDate>Wed, 12 Nov 2025 22:45:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/46cef636-e63a-4011-8660-d03fe8ad8f61.mp3" length="76917090" type="audio/mpeg"/><itunes:duration>53:20</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>full</itunes:episodeType><itunes:season>1</itunes:season><itunes:episode>1</itunes:episode><podcast:episode>1</podcast:episode><podcast:season>1</podcast:season><podcast:transcript url="https://transcripts.captivate.fm/transcript/1063dafc-4f77-4fc0-acc1-49ff1f109081/index.html" type="text/html"/></item><item><title>Chaos Agents Trailer</title><itunes:title>Chaos Agents Trailer</itunes:title><description><![CDATA[<p><em>Chaos Agents</em> is the AI podcast where technologists <strong>Sara Chipps</strong> and <strong>Becca Lewy</strong> try to make sense of a world moving faster than ever. Each week, they dive into the wild, funny, and sometimes weird frontier of <strong>artificial intelligence, technology, and culture</strong>—how it works, what it means, and why it matters. From coding with AI and open-source revolutions to the ethics, creativity, and chaos reshaping our future, Sara and Becca learn out loud with brilliant guests (and the occasional chatbot). Smart, curious, and a little chaotic, <em>Chaos Agents</em> is your invitation to laugh, learn, and keep up with the machines.</p>]]></description><content:encoded><![CDATA[<p><em>Chaos Agents</em> is the AI podcast where technologists <strong>Sara Chipps</strong> and <strong>Becca Lewy</strong> try to make sense of a world moving faster than ever. Each week, they dive into the wild, funny, and sometimes weird frontier of <strong>artificial intelligence, technology, and culture</strong>—how it works, what it means, and why it matters. From coding with AI and open-source revolutions to the ethics, creativity, and chaos reshaping our future, Sara and Becca learn out loud with brilliant guests (and the occasional chatbot). Smart, curious, and a little chaotic, <em>Chaos Agents</em> is your invitation to laugh, learn, and keep up with the machines.</p>]]></content:encoded><link><![CDATA[https://chaosagents.ai]]></link><guid isPermaLink="false">eddfa79f-c1c9-4567-9c68-20d1a4002b14</guid><itunes:image href="https://artwork.captivate.fm/11c6e333-9c74-4a65-a394-8a78b1947cef/headshot.jpg"/><pubDate>Sun, 02 Nov 2025 16:27:00 -0500</pubDate><enclosure url="https://episodes.captivate.fm/episode/eddfa79f-c1c9-4567-9c68-20d1a4002b14.mp3" length="3719859" type="audio/mpeg"/><itunes:duration>02:33</itunes:duration><itunes:explicit>false</itunes:explicit><itunes:episodeType>trailer</itunes:episodeType><podcast:transcript url="https://transcripts.captivate.fm/transcript/6bd882d3-d6ea-48c0-b0df-c38af40f4c5d/index.html" type="text/html"/></item></channel></rss>