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

Technology

Podcast website

myweirdprompts.com

Latest episode

Nov 21, 2026

Where to listen?

Podcasts in the app Replaio Radio Coming soon

Podcasts are coming to the app soon. Install now and be the first to see a whole new take on podcasts

Get it on Google Play Install for free Android 10M+ downloads · 4.7+ rating iOS soon

Episodes

Strip Your Agent to Bash 12.04.2026

LangGraph, CrewAI, AutoGen, Semantic Kernel, Claude Code—they all orchestrate LLM calls with tools, but they encode radically different philosophies about how agents should operate. This episode digs into what actually distinguishes one agentic framework from another, and why the real engineering creativity lives in the harness, not the model. We walk through concrete data: how Vercel deleted 80%...

Why Bigger Context Windows Don't Fix Attention 12.04.2026

A landmark study replaced irrelevant tokens with blanks—and reasoning still collapsed. This episode explores why context degradation is baked into attention mechanics, not retrieval, and what that means for long-horizon AI reasoning. Episode #748509 — open it directly at myweirdprompts.com/748509

Designing Autonomy Boundaries for AI Agents 12.04.2026

When do AI agents actually need to pick their own tools? Daniel's question digs into the spectrum from fully autonomous tool selection (AutoGPT, MCP servers) to deterministic orchestration (LangGraph, CrewAI, Bedrock). The answer isn't about safety blankets—it's about token economics, the Context-Capability Paradox, and what production deployments actually reveal about where autonomous agents fail...

When Knowledge Work Stops Being Safe 12.04.2026

For sixty years, the knowledge economy was supposed to be the safe harbor from automation. Get educated, become a consultant or analyst, and you'd be protected. That deal held until November 2022. This episode traces three eras of labor history—the Industrial Era, the Knowledge Economy Era, and what's happening now—to understand why knowledge workers thought they were untouchable, and why current...

Claude's Latency Profile and SLA Guarantees 12.04.2026

When developers say Claude is slow, what do they actually mean? This episode digs into the five core latency metrics that matter for production systems, reveals the benchmarks showing Claude's p95 latency problem, and then explores what Anthropic actually contractually guarantees—spoiler: almost nothing at standard tier. We break down Priority Tier's queue-prioritization illusion, why Fast Mode's...

When the State Protects Politicians, Not People 12.04.2026

What happens when a government delivers security theater instead of actual security? After six weeks of Iranian missile fire, Israel's ceasefire left its stated military aims largely unmet—Iran retains enriched uranium and can rebuild its missile capability. But the deeper crisis isn't military: it's political. While citizens sheltered nightly with children, the government passed a wartime budget...

Claude Managed Agents: Brain Versus Hands 11.04.2026

Anthropic launched Claude Managed Agents in public beta on April 8th, positioning it as a hosted execution runtime for agentic workflows. Unlike OpenAI's Assistants API—which was primarily a state management layer—Managed Agents includes a real Linux container sandbox, persistent sessions, multi-agent coordination, and governance features like scoped permissions and execution tracing. But the trad...

Do You Become More You? 11.04.2026

What makes you you? This episode explores the science of personality formation, from the Big Five model to twin studies and the Dunedin study. We examine how genetics and environment interact, why early childhood temperament predicts adult outcomes, and why the "personality sets by 30" myth is wrong. Learn how personality actually changes across your lifespan—and why the same parents can raise ver...

The Architecture of Opacity in Think Tank Funding 11.04.2026

When think tank funding is designed to be invisible, how do you evaluate credibility? This episode examines the sophisticated financial plumbing that launders foreign influence into policy research, and why the problem is structural, not just a matter of disclosure. Episode #495462 — open it directly at myweirdprompts.com/495462

Public Affairs vs. Lobbying: Shaping the Battlefield 11.04.2026

What do public affairs firms actually do? It’s more than just lobbying. We explore how these firms shape policy outcomes by managing an organization's entire political and social environment. From legislative tracking software like FiscalNote to geopolitical risk modeling, public affairs is the operating system, while lobbying is just one application. We examine how firms navigate the collision of...

Iran's Shadow Architecture Beyond Missiles 11.04.2026

Most coverage focuses on Iran's military proxies, but a deeper shadow architecture drives its influence. This episode explores the financial networks, religious institutions, and diplomatic maneuvering through BRICS and the Shanghai Cooperation Organization that sustain Iran's power. Learn how Tehran is building a sanction-proof financial corridor and embedding itself in alternative international...

How Lobbying Actually Works in DC 11.04.2026

Federal lobbying spending surged to $6 billion in 2025, a 36% jump driven by debates over AI regulation, trade tariffs, and healthcare policy. This episode breaks down what lobbying actually is—from the "information subsidy" lobbyists provide to the granular data models they use to influence lawmakers. We explore the daily reality of the job (it's more administrative than martini lunches), the rev...

A Baby's Mouth Is a Lab-Grade Sensor 11.04.2026

When a baby starts crawling, the entire house becomes a sensory buffet, and the mouth becomes a high-resolution 3D scanner. This episode explores the developmental science behind why babies explore with their mouths and offers a practical framework for parents to evaluate household objects. Learn to distinguish between mechanical choking hazards and chemical risks, and discover how to curate a "Ye...

The Minefield of Information 11.04.2026

A ceasefire is declared, but the fighting rages on. Diplomats meet in Islamabad while the Strait of Hormuz is choked by strategic ambiguity and an information blackout. We break down the contradictions of the Iran conflict: why Iran claims it can't find its own mines, how Trump's Truth Social posts become negotiation leverage, and why the information void is doing active work. From OODA loops to P...

Debugging Your Brain’s Source Code 11.04.2026

In this episode, we break down a powerful cognitive framework called "The Model," which deconstructs every emotional reaction into a five-step causal chain: Circumstance, Thought, Feeling, Action, and Result. We explore how this sequence acts like a debugger for the brain, revealing that neutral events don't cause our feelings—our interpretations do. By treating thoughts as optional code rather th...

Pakistan's Two-Track Diplomacy 11.04.2026

Why would Pakistan host high-stakes US-Iran peace negotiations in Islamabad just days after its Defense Minister publicly condemned Israel? This episode explores the complex motivations behind Pakistan's role as a mediator, revealing how border security, energy needs, and diplomatic prestige align to make Pakistan a uniquely credible broker. We examine the ISI's parallel intelligence channels with...

IRGC: From Street Militia to Regional Franchise 10.04.2026

In this episode, we unpack the IRGC's transformation from a ragtag revolutionary guard into a sophisticated "franchise" model for regional influence. We explore the ideological seeds planted in 1979, the economic engine that funded their expansion, and the "advisory" playbook used to build proxy states. From the Bekaa Valley to the rise of the Popular Mobilization Forces in Iraq, we reveal how the...

Israel's New Axis: Beyond Washington 10.04.2026

Everyone looks at the Middle East map and sees the United States as the obvious cornerstone of Israel's defense. But look closer at the data from 2024 through 2026, and a different story emerges: a quiet, stealthy consolidation of a new axis of support stretching from the Persian Gulf to the Bay of Bengal. While traditional European capitals are becoming fair-weather friends, the real action is mo...

The AI Wargame's Flat Hierarchy Problem 10.04.2026

The promise of AI in geopolitical wargaming is simulating thousands of perspectives simultaneously. But there's a critical flaw: Large Language Models treat every actor as a peer, giving equal weight to a press release from a local NGO and a troop mobilization order from a superpower. This episode explores the "Exhaustive List Fallacy," why adding more actors often makes simulations less accurate,...

Why Skin in the Game Beats Silicon in Forecasting 10.04.2026

Prediction markets, expert elicitation, and AI wargames all claim to predict the Iran-Israel ceasefire. This episode examines why financial incentives often outperform pure simulation, and how to triangulate signals when the world is on fire. Episode #852160 — open it directly at myweirdprompts.com/852160

AI Wargaming: One Model or Many? 10.04.2026

Should geopolitical AI simulations use one model or many? We debate the pros and cons of a single-model approach. This episode explores the tension between scientific control and real-world fidelity in AI wargaming. Episode #414118 — open it directly at myweirdprompts.com/414118

Simulating Geopolitics Under Asymmetric Information 09.04.2026

How a two-stage AI pipeline—combining Monte Carlo simulation with an LLM council—models ceasefire durability by forcing actors to make decisions with incomplete information. This episode unpacks the architecture, the divergences between stages, and what the method reveals about live crisis forecasting. Episode #476510 — open it directly at myweirdprompts.com/476510

The Nervous System of Multi-Agent Systems 09.04.2026

How do parent agents know when their subagents are truly done? We trace the lifecycle management, state transitions, and coordination overhead that turn AI chat into distributed computing. Episode #773469 — open it directly at myweirdprompts.com/773469

Choosing Your Durable Execution Platform 09.04.2026

Facing the infrastructure tax of building AI agents? We compare Temporal, AWS Step Functions, Google Cloud Workflows, and Azure Durable Functions to find the best fit for your code-defined agentic workflows. Episode #402254 — open it directly at myweirdprompts.com/402254

A Functional Chaos: Middle East 2027 09.04.2026

One year after the Great Reset, the Middle East map has been redrawn. In this episode, we look ahead to April 2027 to predict the long-term fallout of the May 2025 conflict. We explore the collapse of the singular Supreme Leader in Iran, replaced by the factional "Council of Five." We track Israel's massive migration of its tech economy to the Negev desert, creating a "garrison tech state." Plus,...

Listen to the My Weird Prompts podcast in Replaio

Radio and podcasts in one app - free, with no sign-up. Install today and do not miss the launch

Get it on Google Play

Replaio is not a podcast publisher; show names, artwork and audio belong to their authors and are distributed through public RSS feeds.