Pilot — Second Brain for AI, with George B. Thomas
George B. Thomas joins Chris Carolan for the pilot episode of The AI-Native Shift to unpack why AI feels amnesic and how a markdown-based "second brain" turns AI from a stateless conversation partner into a persistent memory layer. They explore George's "folders with files with rules" architecture, the case for plain markdown over vendor-locked tools, and what unlocks when a non-developer treats Claude Code as the implementation layer.
Episode aired 2026-05-27. George went deep on his 'main brain' architecture (his deliberate term — not 'second brain') built on Obsidian + Claude Code with local Gemma 3.12 as a token-cheap helper. He demoed live: Echo (his conversational assistant) speaking back unprompted via TTS, the Main Brain Command Center (an Obsidian plugin he wrote in one Saturday morning), and the folder discipline of raw/wiki/intake/journal that lets the system 'compound' over months. Chris closed with the show's working thesis: 'show and tell building stuff... what it looks like to be AI native.'
Show Notes
Key Topics Covered
- ✓ Main brain naming distinction from second brain (main brain is in charge of knowledge, not auxiliary to it)
- ✓ Folder discipline: raw (append-only) / wiki (atomic linked pages) / intake (inbox) / journal (auto-written from sessions)
- ✓ Identities as a spice rack — per-context voice profiles swappable per task (training vs writing vs videos vs per-client)
- ✓ Slash commands as system primitives: /intake process, /dream (end-of-day optimizer), /save context, /recall
- ✓ Main Brain Command Center: Obsidian plugin with live integrations (Gmail, Calendar, Zoom, ClickUp, Slack, HubSpot) and TTS voice
Enjoying the Show?
Subscribe to The AI-Native Shift and never miss an episode.