Luminary Life
The Builders Mastermind
The Luminary Life front door and member platform: an AI-vetted application, invitation-to-membership billing, an events engine, and a Ship Log where members post what they ship.
01 The story
The public face of Luminary Life and the platform its members work inside. Applications flow through a segmented form, a server computes the tier honestly, and an AI vetting agent recommends fit with a sanitized rationale; a securities and FTC banned-phrase list fails the production build if any forbidden claim appears anywhere. In July the member side was rebuilt in a single 26-hour push: eight surfaces in place of a thirteen-item menu, an invitation flow built to carry an accepted applicant through Stripe checkout to a linked account and welcome email (wired, awaiting activation), an events engine with RSVP, waitlist, magic-link entry, and check-in, and a Ship Log where members declare a build and post wins to the room. Lumen runs inside it on a per-member monthly spend ceiling, so AI cost can never run away quietly. The whole platform installs as an app and keeps a working shell offline.
02 What went into it
- Four-track segmented application with local draft autosave
- Server-side tier computation and AI vetting with sanitized output
- Build-failing banned-phrase compliance gate
- Invitation to membership: forward-only state machine, Stripe checkout, idempotent webhook, account link and welcome email
- Events engine: listing, recap mode for search, RSVP with capacity and waitlist, magic-link entry, calendar export, lifecycle-guarded admin
- Ship Log: declare a build, post ships and wins to the room, owner-only writes enforced in the database
- Four briefing channels on 20 vetted feeds, every bullet tied to a dated source
- Lumen chat under a per-member monthly AI spend ceiling with an admin cost view
- Referral pipeline visibility scoped to the member's own code
- Installable PWA: offline shell, versioned caches, never caches auth or API responses
- One signed-cookie gate helper, fail-closed, shared by every private room
03 Claude in production
Every AI call runs server-side under the Lumen governance layer: no client keys, cost telemetry on every call, and a human confirming every consequential write.
04 Stack and capabilities