A private second brain, a self-querying knowledge graph, a command center, the automation that ties it together, and a public front door that turns conversations into leads.
Private data stays on the machine. The shareable Brain vault sits on top of it, a knowledge graph makes it queryable, local models do the private heavy-lifting, the Command Center runs the day, and remote access reaches it all from anywhere.
The six layers are the private, local stack. callconnor.com faces outward: an "Ask Connor" chatbot that talks to visitors, researches their company live, scores the conversation, and drops a qualified lead into the pipeline.
The same file-based, transparent-scoring patterns aimed at one job: a Partner Success console for a brokerage's accounts. A transparent engine scores each one in context, then surfaces the one play that moves it.
The single line that lets the whole vault be pushed to GitHub without worry: raw personal data never leaves the machine, and only a scrubbed, human-reviewed summary ever crosses into Brain.
Finances and Relationships never cross. Ever. Health stays raw in the Journal; only a thin overview reaches Brain. The private side runs its own local-model advisors instead — nothing has to cross to be useful.
Same tool (graphify), two independent corpora, exposed to Claude as read-only MCP servers.
Two schedulers on purpose: macOS launchd for OS jobs, Claude's own scheduler for skill runs.
The core landed in a single May sprint. Everything since is deliberate accretion: new systems added weekly, each verified before it ships. 405 commits in the Brain repo and counting.