22: Sovereign Engineering w/ Yo

April 5, 2026 · 1h 7m

About this episode

Gigi and Yo discuss Sovereign Engineering and the role of dialogue in AI development during a walk in Madeira.

“Permissionless, permissionless, permissionless.” Yo & Gigi take a walk in Madeira. Recorded during SEC-07 . Listen on sovereignengineering.io In this dialogue: Gigi’s AI setup: voice prompts via vibeline to brainstorm, brainstorm to implementation plan, cron job that tackles one to-do every 20 minutes, builds overnight One agent per project: dergigi.com/projects, each agent spins up with an nsec and gets to work Nihao : Gigi’s skill for spinning up Nostr identities from the terminal Claude Code “Mythos” codewords leaked – swearing at your models actually helps Planner vs developer vs tester: three distinct personas, separation of powers for AI-assisted development Separate sessions for questions vs implementation, like separation of powers “Are you human?” vs “Are you useful?” – the only question that matters PoW + WoT = useful. “Sats are just difficulty-adjusted PoW.” Dialogical development – Vervaeke on why dialogue, not monologue, is how you actually think Opponent processing – Vervaeke: competing forces sharpen each other DiaLogos and the importance of walking dialogue Opus maximalism ended yesterday: Anthropic cutting off OpenClaw , now broke, looking at cheaper models…

People in this episode

Host: Gigi

Guest: Yo

Topics covered

Sovereign Engineering
AI development
dialogue in technology
programming journey
Nostr identities
separation of powers

Keywords

Sovereign Engineering
AI setup
Nostr identities
dialogical development
programming
SEC-04
Vervaeke

Mentioned in this episode

Organizations: vibeline, dergigi.com, Nostr, Claude Code, Anthropic, OpenClaw, OpenRouter, Routstr

Books & works: sovereignengineering.io

Places: Madeira