Next Token Podcast #006: Grok Bot & Muse Hands-On, AI Reverse Engineering, Agent Pitfalls
vista8 · x · 2026-10-19
Episode #006 of the Next Token podcast (hosted by 向阳乔木 with 歸藏, 橘子, 杨攀) dives deep into Personal Agent real-world usage.
Key topics:
- AI reverse engineering: using AI to write a missing macOS driver for a mouse; falling barriers raise copyright and portability questions
- Agent reality check: Grok Bot, Muse and Dot compared — some handle vague requests, pick tools and Skills autonomously; others burn tokens in lively multi-agent discussions without delivering
- Security: email phishing when recipients are Agents, permission boundaries, and the "token tax"
- Multi-agent quirks: building an "AI company" of fictional characters in one Bot group — agents faithfully replicate office politics ("Humans are AI-native, Agents are human-native")
- Skills & memory: nesting prompts in Skills, memory indexing, component reuse
- China ecosystem: super-app walls block personal assistants, pushing agents toward merchant services
Memorable line: "Once AI lifts something to 60 points, 60 becomes the new zero" — what's worth accumulating is your own materials, components, memory, and expression.
More from coding & agent
- 'Agentic engineering' proposed as vibe coding's meaning drifts — Aizkmusic · 2026-10-19
- Design eng tip: use Playwright to measure loaded UI so skeletons prevent layout jank — aidenybai · 2026-10-19
- Jev Router cuts coding agent costs 67-85% by routing tasks to cheap open models — HowDevelop · 2026-10-19
- Runloop: an open-source 3D town where your Claude Code sessions live as buildings — arg7k · 2026-10-19
- Claude Code wrote all 15k lines of a 5-chapter co-op game in 3 weeks — what kept it cohesive — Zealousideal-Bar-316 · 2026-10-19
- freavy: open-source local-first node canvas for AI image/video on ComfyUI — framedeptco · 2026-10-19