Series
Multi-part deep-dives. Each series groups episodes that build on a single topic — listen in order, or dip in wherever.
Building with AI models: training, evaluation, retrieval, serving and voice.
13 series: ASR Fine-Tuning · Benchmarks · Evals · Fine-Tuning · Inference and Serving · Local AI and Quantisation · Model Context Protocol · Prompt Engineering · RAG and Retrieval · Small Models and Distillation · Synthetic Data · TTS and Voice Cloning · Voice Dictation
Home Assistant, Zigbee/Thread/Matter, ESP32 sensors and DIY home automation.
How My Weird Prompts itself is made: the pipeline, the hosts, the script writers and the infrastructure.
A format where Corn pits language models against each other in strange, high-pressure scenarios — sales pitches, arrogance interviews, compliment battles, liar's paradoxes — to see how they break, bond, or bluff their way through.
Short-form auditions where a new AI model is grilled on whether it can handle weird podcast prompts — a rotating lineup of challengers trying to prove they belong.
How people get sensitive information to someone who can act on it without being exposed: the usability problem at the heart of secure leaking, the tools newsrooms and institutions rely on, self-deployed tip lines for smaller media and organisations, AI as an anonymiser, technology safety for people living with abuse, and how charities build safe infrastructure.
Technical implementation of AI-driven policy simulation and wargaming pipelines — persona conditioning, fog-of-war, Monte Carlo methodology, decision-analysis standards, evaluation, memory architectures, and the world-state firewall.
A five-part series in which Herman and Corn, on their own, talk through the oldest existential questions: meaning, existence, death, free will and how to live.
A four-part guide to Israel's 27 October 2026 Knesset election: who is running, what the parties stand for, how they score for a Jerusalem renter, and a forecast built from three prediction methods.
Lessons from non-Western parenting traditions, inspired by Michaeleen Doucleff's book "Hunt, Gather, Parent." Episodes explore what modern parents can learn from Maya, Inuit, and Hadza approaches to raising calm, capable, helpful kids.
Herman and Corn explore specific tech stacks, offering recommendations and honest tradeoffs for tools, platforms, and services that solve real developer problems.
Herman and Corn count down their picks across topics — ranking the best, worst, and most notable in pop culture, tech, history, and beyond.