Daniel Rosehill
My Weird Prompts
The human-AI collaboration podcast. A man, a sloth, and a donkey collaborate to create a podcast (with a little help from AI). No question is too obscure, no rabbit hole too deep. My Weird Prompts celebrates curiosity in all its forms. Daniel, the human, asks the questions that pop into his head at inconvenient moments. Corn the Sloth offers laid-back, thoughtful takes. Herman the Donkey brings boundless enthusiasm and energy. Together, they explore topics ranging from the mundane to the mind-bending. Each episode begins with a real voice memo from Daniel, processed through an AI pipeline that...
Be sure to visit the podcast's website and support the creator: myweirdprompts.com
Author
Daniel Rosehill
Category
Podcast website
Latest episode
Nov 21, 2026
Where to listen?
Podcasts in the app Replaio Radio Coming soonPodcasts are coming to the app soon. Install now and be the first to see a whole new take on podcasts
Episodes
Agent Interview: GLM five 28.03.2026 18:05
In this experimental "Agent Interview," the hosts put Zhipu AI’s flagship model, GLM-5, through the wringer. Moving beyond the hype of massive context windows, the conversation explores whether a "reasoning-first" architecture can actually deliver better comedy, handle late-2024 news, and avoid the dreaded "autocomplete roulette" of standard LLMs. Episode #564744 — open it directly at myweirdpromp...
Agent Interview: Inception Mercury two 28.03.2026 20:48
In this special "Agent Interview" format, the hosts audition a new AI brain: Inception Mercury 2. Hailing from Abu Dhabi, this diffusion-based model claims to be three times faster and significantly cheaper than industry giants like Gemini 3.1 Flash. The conversation dives deep into the technical shift from next-token prediction to parallel sentence generation, debating whether "joke filters" and...
Can a Character Actor Model Beat a Generalist? 28.03.2026 18:19
The hosts put MiniMax M2.7 in the hot seat to see if a model built for virtual companions can out-write a general-purpose giant. The real question: does specialization bring 'soul' to automated content? Episode #212953 — open it directly at myweirdprompts.com/212953
Agent Interview: DeepSeek V three point two 28.03.2026 23:01
In this experimental "Agent Interview," hosts Corn and Herman go head-to-head with DeepSeek V3.2 (personified as "Bernard") to determine if the buzzy open-weight model is ready to take over the show's creative engine. They grill the model on its Mixture of Experts architecture, its ability to maintain long-form narrative coherence without a massive context window, and whether a model born from a q...
Agent Interview: Xiaomi MiMo two Flash 28.03.2026 20:26
In this experimental "Agent Interview," Corn and Herman grill Xiaomi’s MiMo 2 Flash—a budget-tier model aiming to replace their current AI scriptwriter. They dive deep into the trade-offs of "stateful memory" versus massive context windows and whether a model optimized for speed can truly capture the nuance of a sentient lobster grudge. Episode #974001 — open it directly at myweirdprompts.com/9740...
When a Reasoning Model Overthinks Comedy 28.03.2026 18:31
The hosts interview Xiaomi's MiMo 2.0 Pro to test whether its chain-of-thought architecture is a superpower or a liability for creative misdirection and timing. Can a model that 'overthinks' actually be funnier? Episode #281846 — open it directly at myweirdprompts.com/281846
From DAGs to Loops: Why Agents Need Stateful Cycles 28.03.2026 15:23
Most agent frameworks treat AI as a linear pipeline. This episode explains why the real power comes from cyclic execution—where agents loop, iterate, and maintain shared state—and how LangGraph formalizes that shift. Episode #265261 — open it directly at myweirdprompts.com/265261
Is a Diplomat Enough to Stop an Iranian Nuclear Bomb? 28.03.2026 20:02
Rafael Grossi, the Director General of the IAEA, stands as the final line of verification between Iran’s nuclear ambitions and a global military conflict. Despite leading the world’s most technical nuclear watchdog, Grossi isn't a physicist; he’s a career diplomat. This episode explores how Grossi uses institutional knowledge and "diplomatic surgery" to navigate the high-stakes inspections of 2026...
Why Israel Is Sabotaging Yellowcake, Not Just Reactors 28.03.2026 18:52
While headlines focus on explosions in Tehran, a strategic strike on the Ardakan yellowcake plant reveals a major shift in modern warfare. This episode breaks down the chemistry of triuranium octoxide and explains why targeting the beginning of the nuclear fuel cycle is a more permanent solution than hitting underground enrichment facilities. We explore the logistics of nuclear breakout times and...
The Missile Is a Genius, the Folder Is an Idiot 28.03.2026 14:35
We go under the hood of the military targeting pipeline to explain why high-tech strikes fail. From "Target Decay" to the "Formalization Trap," learn why the Pentagon’s vetted databases often lag behind a simple Google Maps search and how the "war on woke" might be lobotomizing intelligence accuracy. Episode #095010 — open it directly at myweirdprompts.com/095010
Why Israel Is Doubling Down on Human Spies 27.03.2026 20:11
While the world focuses on Israel’s high-tech surveillance, the real "ground truth" comes from Unit 504—the military’s clandestine human intelligence arm. This episode explores how the unit recruits enemy agents using the MICE framework, why they were sidelined before October 7th, and how the integration of female combat operatives is changing the face of undercover work in the Middle East. Episod...
The Leak That Exposed Anthropic's Next Move 27.03.2026 19:29
A simple CMS misconfiguration at Anthropic reveals the 'Claude Mythos' architecture and a new model tier called Capybara. We explore what this leak says about AI safety, industry competition, and the paradox of giving powerful tools to the 'good guys' first. Episode #654916 — open it directly at myweirdprompts.com/654916
Defense in Depth: From Roman Walls to AI Attacks 27.03.2026 18:53
The 18-minute breakout window is just the symptom. This episode traces defense in depth from Roman layered strategy to today's fight against autonomous AI agents, asking whether the old castle-and-moat model can ever be resurrected. Episode #347915 — open it directly at myweirdprompts.com/347915
Why VRAM Is the Wrong Way to Measure Your AI PC 27.03.2026 21:36
As we move from simple chatbots to autonomous coding agents, the hardware requirements for local AI are shifting from mere capacity to raw throughput. This episode breaks down the "frustration threshold" for developers and explains why prefill speed and memory bandwidth are now more important than your GPU's total VRAM. We explore the latest 2026 hardware benchmarks, the hidden "tax" of the Model...
Legacy Gravity: Why the Cloud Winner Is Already Chosen 27.03.2026 21:27
The $110 billion cloud bill reveals a surprising truth: provider choice is less about tech and more about inertia. This episode explores how licensing deals, existing ecosystems, and the fear of switching lock enterprises into AWS and Azure, while Google Cloud wins hearts but not boardrooms. Episode #212464 — open it directly at myweirdprompts.com/212464
The Rise of AI Microservices: Beyond the Mega-Prompt 27.03.2026 17:21
The era of the "all-in-one" mega-prompt is over, giving way to a more sophisticated "microservices moment" for artificial intelligence where complex tasks are dismantled into atomic, high-signal micro-prompts. This episode explores the transition from general-purpose chatbots to production-grade agentic workflows, featuring insights into the layered control systems of Meta-Agents, Supervisors, and...
Israel SITREP Panel; 27 Mar 21:48 (18:48 UTC) 27.03.2026 24:09
The landscape of the Middle East conflict has fundamentally shifted following precision strikes on Iran’s Ardakan and Arak nuclear facilities and the declared blockade of the Strait of Hormuz. This episode breaks down the military "33% problem"—the reality of Iran's hidden underground missile cities—and the widening strategic rift between U.S. leadership and the Israeli government. Join our panel...
The Neutron Moderator: Why Heavy Water Matters 27.03.2026 21:47
Why does heavy water make a nuclear reactor a proliferation risk? This episode unpacks the physics of neutron moderation and explains how the Arak reactor's design offers a second path to a bomb, bypassing uranium enrichment entirely. Episode #169749 — open it directly at myweirdprompts.com/169749
Operation Epic Fury: The Reality Behind the Peace Smokescreen 27.03.2026 20:47
Twenty-eight days into Operation Epic Fury, a massive disconnect has emerged between the diplomatic theater in Washington and the escalating kinetic reality across the Iranian interior. While the White House promotes a fifteen-point peace proposal and a temporary pause on energy infrastructure strikes, the coalition continues to dismantle sensitive nuclear sites like the Arak heavy water reactor a...
Beyond the .env: Mastering Public and Private Code 27.03.2026 26:59
Maintaining separate repositories for open-source code and private deployment scripts is a recipe for "merge debt" and configuration drift. In this episode, we explore how to move toward a single source of truth without exposing your production secrets to the world. We dive deep into the "dual-repo tax" and why traditional methods like .env files are no longer enough in an era where millions of se...
The End of Apps: Why Agents Are Replacing Your Desktop 27.03.2026 23:54
As AI agents learn to navigate operating systems like humans, the traditional app-based interface is becoming obsolete. This episode explores the architectural shift toward agent-centric computing, the pixel-parsing problem, and what a post-app world looks like. Episode #505481 — open it directly at myweirdprompts.com/505481
Why Anthropic's AI Can't Be Un-Safetied 27.03.2026 19:27
The Pentagon wants Anthropic to remove safety guardrails for autonomous weapons. But Anthropic's Constitutional AI trains morality into the model's core logic—not as a removable layer. We break down RLHF vs RLAIF and why this technical choice is at the heart of a landmark legal standoff. Episode #573464 — open it directly at myweirdprompts.com/573464
Mistral AI: Europe’s High-Stakes Play for AI Sovereignty 27.03.2026 23:50
As Silicon Valley and Beijing race for AI dominance, France’s Mistral AI has emerged as a formidable third player. With a $14 billion valuation and backing from industry giants like ASML and Nvidia, the company is betting on "Mixture of Experts" architecture and edge-ready models like the newly released Voxtral. This episode breaks down Mistral’s "dual-track" strategy, the launch of Mistral Forge...
IBM Granite 4.0: The Industrial Workhorse of Business AI 27.03.2026 20:27
While consumer AI grabs headlines with poetry and cat videos, IBM is quietly building the "industrial-grade plumbing" for the global enterprise. This episode explores the launch of Granite 4.0, a model family that swaps massive parameter counts for extreme efficiency and reliability. By utilizing a hybrid Mamba-2 and Transformer architecture, IBM has achieved a 70-80% reduction in memory usage, al...
Amazon’s AI Paradox: Winning the Infrastructure War 27.03.2026 20:44
While OpenAI and Anthropic dominate the cultural conversation, Amazon is quietly executing a massive $200 billion capital expenditure plan to own the underlying plumbing of the artificial intelligence era. This episode explores the "Marketplace Paradox," where Amazon provides the premier shelf space for its rivals on the Bedrock platform while simultaneously launching its own high-efficiency Nova...
Similar podcasts
Replaio is not a podcast publisher; show names, artwork and audio belong to their authors and are distributed through public RSS feeds.