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
NVIDIA’s $26 Billion Pivot: From Chips to AI Models 27.03.2026 18:35
For years, NVIDIA has been the undisputed king of AI hardware, but a massive shift is underway. This episode dives into the recent GTC announcements, where the company unveiled the Rubin platform, the Vera CPU, and a staggering $26 billion push into open-weight models like the Nemotron series. We explore how vertical integration—combining custom silicon with specialized AI intelligence—is creating...
DeepSeek’s Return: V4, R2, and the AI Pricing War 27.03.2026 22:48
After a year of silence, DeepSeek has returned to the spotlight with the launch of V4 and R2, sending shockwaves through the AI industry with a trillion-parameter architecture and unprecedented pricing. This episode dives into the technical breakthroughs of Manifold-Constrained Hyper-Connections and Mixture of Experts that allow such a massive model to run with incredible efficiency on domestic Ch...
Alibaba’s Qwen 3.5: The New King of Intelligence Density 27.03.2026 17:57
Alibaba has sent shockwaves through the AI industry with the release of the Qwen 3.5 series, proving that size isn't everything when it comes to reasoning. This episode explores the concept of "intelligence density," where a 9-billion parameter model is outperforming Western giants on graduate-level science benchmarks. We dive into Alibaba's aggressive "Model-as-a-Service" strategy, which aims to...
The $3 Billion Stealth Giant: AI21 Labs & Nvidia 27.03.2026 20:30
As reports surface of a potential $3 billion acquisition by Nvidia, we dive into the story of AI21 Labs, the Israeli powerhouse that has spent years building the "plumbing" of the AI revolution. While others chased viral chatbots, AI21 focused on enterprise-grade reliability and architectural innovation, culminating in the groundbreaking Jamba model. This episode explores how their hybrid Mamba-Tr...
Fire Your Software Subscriptions and Just Code the Vibe 27.03.2026 24:13
Stop renting your productivity and start owning it. This episode explores the "subscription graveyard" and the revolutionary shift toward "vibe coding," where non-technical users leverage agentic workflows to build custom, self-healing tools in hours rather than months. From fixing niche Hebrew formatting issues to replacing bloated CRMs, we discuss how the 85% drop in API costs is dismantling the...
When AI Truth-Seeking Meets the Law 27.03.2026 21:02
Grok 4.20's multi-agent architecture aims for unfiltered truth, but a Dutch court's daily fines for deepfakes reveal the growing friction between raw computational power and global regulation. How far can agentic AI push before the law pushes back? Episode #178079 — open it directly at myweirdprompts.com/178079
Cohere: The Switzerland of Enterprise AI 27.03.2026 18:26
While consumer-facing chatbots dominate the headlines, Cohere is methodically building the high-stakes infrastructure for the modern enterprise. Dubbed the "Switzerland of AI," the company has carved out a unique position by remaining cloud-agnostic and focusing on the unglamorous but essential needs of banks, healthcare systems, and defense contractors. This episode examines Cohere’s strategic fo...
The Digital Tofu Crisis: Saving the World’s Scripts 27.03.2026 18:59
While our digital devices are packed with thousands of emojis, nearly half of the world’s writing systems remain digitally invisible, often appearing only as empty boxes known as "tofu." This episode dives into the technical and bureaucratic hurdles of the Unicode Standard, exploring why ancient hieroglyphs and modern minority scripts struggle to gain a foothold in our global digital infrastructur...
Xiaomi's Ghost Model: How Anonymous Testing Built an AI Empire 27.03.2026 17:45
When a mysterious AI model called Hunter Alpha topped the rankings, no one knew it was Xiaomi. This episode explores how the hardware giant used guerrilla marketing and technical brilliance to launch its Agent Era, challenging the AI establishment. Episode #226248 — open it directly at myweirdprompts.com/226248
The Battery Bottleneck: Why Your Phone Still Dies by 10 PM 27.03.2026 21:19
We live in an era of folding screens and two-nanometer chips, yet the average smartphone user remains tethered to a wall every evening. This episode dives deep into the electrochemical and physical bottlenecks preventing smartphone battery density from scaling alongside our processing power. We explore the "Smartphone Envelope," where batteries must compete for precious millimeters against massive...
Why AI Teams Are Hiring Digital Middle Managers 27.03.2026 20:27
The "honeymoon phase" of agentic AI is over. Recent research shows that simply throwing more agents at a problem causes systems to collapse under a "coordination depth wall." To solve this, developers are building something we once tried to escape: bureaucracy. This episode explores the transition from flat orchestrators to sophisticated hierarchical structures like the HiMAC framework. We dive in...
Why Your AI Agent Needs a Ticket System, Not a Chatbox 27.03.2026 17:31
Are your AI agents losing the thread the moment you give them a mid-task instruction? In this episode, we explore the "interruption problem" and why the era of intuitive "vibe coding" is officially over, giving way to a new age of robust agent orchestration. We break down the latest breakthroughs from March 2026, including OpenAI’s Responses API with context compaction and Anthropic’s Dispatch too...
Why Your 2026 Smartphone Still Feels Like a Part-Time Job 27.03.2026 19:26
In 2026, mobile hardware has reached incredible heights, yet the software setup process remains a frustrating "empty room" experience. This episode explores the widening gap between Google’s managed cloud services and the needs of power users who demand total control. We dive into the technical bottlenecks of the 25MB backup cap, the controversial new 24-hour waiting period for sideloading, and ho...
The Soul of Your Machine: Automating What Matters 27.03.2026 23:47
What do you lose when your OS crashes? Not just files, but years of personalized tweaks. This episode explores how treating your desktop like a recipe—not a pet—can save your machine's soul, and why dotfiles are the real battleground. Episode #108665 — open it directly at myweirdprompts.com/108665
The Desert That Holds Time and Refugees 27.03.2026 26:38
How does a geological marvel that records 220 million years of Earth's history become a sanctuary for people fleeing modern conflict? This episode explores the strange intersection of deep time, erosion, and the human need for refuge in the Negev's Makhtesh Ramon. Episode #230143 — open it directly at myweirdprompts.com/230143
The Vector Debt Trap: Choosing Embeddings That Last 27.03.2026 23:13
Why your choice of embedding model today could lock you into a costly re-indexing nightmare tomorrow. We explore how Gemini 2's multimodal leap and OpenAI's Matryoshka learning change the calculus for production RAG systems, and how to avoid architectural regret. Episode #494603 — open it directly at myweirdprompts.com/494603
The Brain’s Nightly Power Wash: Cleaning Away Dementia 27.03.2026 25:03
For decades, scientists wondered how the brain disposed of its metabolic waste without a traditional lymphatic system. This episode explores the groundbreaking discovery of the glymphatic system—a nightly "power-wash" that occurs during deep, non-REM sleep. We dive into the mechanical process of how brain cells shrink to let fluid flush out toxic proteins like amyloid-beta and tau, and why the fai...
Why OSINT Maps Outperform the Nightly News 27.03.2026 32:54
As modern conflicts accelerate into twenty-four-second increments, the traditional twenty-four-hour news cycle has become a strategic liability, leaving a "utility gap" that legacy media can no longer bridge. This episode dives deep into the emerging Defense Intelligence Ecosystem, where specialized organizations like the Institute for the Study of War (ISW) and the Alma Research Center are bypass...
When AI Hallucinates History 27.03.2026 24:58
Generative AI can now fill in what 1920s cameras missed, turning archival film into vivid reconstructions. But as we move from restoration to creation, are we uncovering the past or inventing a version that never existed? This episode explores the technical breakthroughs and the haunting ethical line between preservation and hallucination. Episode #976385 — open it directly at myweirdprompts.com/9...
The Architecture of Sleep: Rebuilding Restorative Rest 27.03.2026 21:56
Sleep is often misunderstood as a simple "on-off" switch, but it is actually a complex biological construction known as sleep architecture. This episode explores the "hypnogram"—the intricate map of cycles the brain navigates every night to ensure both physical restoration and emotional processing. From the "power-washing" effects of deep N3 sleep that clears metabolic waste to the high-activity R...
Building a Closed-Loop Hebrew Learning Stack 27.03.2026 24:24
How do you build a specialized AI workflow that turns spoken English into fully vocalized Hebrew and pipes it directly into a spaced repetition system? This episode explores the technical stack—from HeBERT to RTL rendering—needed to close the loop for Semitic language learners. Episode #102571 — open it directly at myweirdprompts.com/102571
The Rocketbook Sunset and the Search for a Clean Erase 27.03.2026 27:32
With Rocketbook shutting down, what replaces it? This episode explores the material science of whiteboard notebooks and why your marker choice matters more than the board itself for AI transcription accuracy. Episode #078119 — open it directly at myweirdprompts.com/078119
The End of the Digital Sandwich: How AI Learns to Speak Your Language 27.03.2026 29:17
AI translation is moving from clunky, language-specific pipelines to omnilingual models that preserve a speaker's unique prosody across hundreds of languages. This episode explores the technical breakthroughs—from universal phonetic manifolds to disentangled representations—that are collapsing the Tower of Babel. Episode #023388 — open it directly at myweirdprompts.com/023388
Beyond Text: How Gemini 1.5 Flash Is Revolutionizing Audio 26.03.2026 23:17
For years, AI has been forced to "read" speech through inaccurate text transcriptions, losing the nuance of tone, emotion, and environment. This episode explores the shift to native multimodality with Google’s Gemini 1.5 Flash, a model that processes raw audio waveforms directly. We break down the technical breakthroughs of the "Audio Haystack" test, the massive million-token context window, and h...
Escaping the Bubble: Building a Better Information Diet 26.03.2026 20:40
In an era of "semantic collapse," major news aggregators often serve a narrow, engagement-driven version of reality. This episode explores why mainstream platforms feel like the "fast food of information" and offers a technical roadmap to a more inclusive, high-signal news diet. We dive into tools like Ground News and AllSides for bias detection, the resurgence of RSS for source control, and decen...
Similar podcasts
Replaio is not a podcast publisher; show names, artwork and audio belong to their authors and are distributed through public RSS feeds.