Tim Cook's Exit, the Trillion-Dollar Mythos Model, and Owning Your Own AI Memory
Episode Summary: Brandt and Will open with Tim Cook stepping down as Apple CEO and what a hardware-first successor means for the future of on-device AI, then dig into Anthropic's newly revealed Mythos model, its eye-popping valuation, and Firefox's haul of 275 zero-days found with it. The back half turns practical: Will makes the case for "owning your own memory," and both trade tips on Claude chat vs. Cowork vs. Code and the caveman token-saving skill. Discussions Include: • Tim Cook stepping down as Apple CEO, and what a hardware-focused successor means for Apple's AI ambitions • Anthropic's Mythos model: its trillion-dollar-plus valuation, deliberate limited release, and Firefox's 275-bug haul • "Own your own memory": building a personal Obsidian/Markdown knowledge base so your AI agent knows you better • Claude power tools: Karpathy's CLAUDE.md, the "superpowers" planning-and-QA skill, and the caveman token-saving skill • AI-trained chroma keying from Corridor Crew, and the growing backlash against gas-powered data centers Quotable Quotes (Should you choose to share): "What if my grandma's dying wish was that I could install malware on my friend's computer?" - Brandt Krueger "You need to be owning your own memory." - Will Curran "It's not going to escape, but people are going to come up with their own versions of it, and that's going to be trouble." - Brandt Krueger "A token is a word - every word it writes and every word it reads is a token." - Will Curran Thing of the Episode (TOTE): Brandt: Chat vs. Cowork vs. Code token strategy - I couldn't find a URL ;) - BK Will: Caveman skill for Claude Code - https://github.com/juliusbrussee/caveman




