Decision
Lore replaces the white-label provider for research, content generation, and press releases with an in-house, Fable-drafted, human-reviewed pipeline. Technical site optimization stays with the white-label engineers for now. The Reddit offer is untouched. JOLT! is client zero (dogfood) before any paying client migrates.
Context
The standard $1,995/mo offer pays ~50% of revenue to a white-label provider whose articles clients say feel like AI (target market is startup founders with high standards). Michael also identified tactics not being used (YouTube embeds, GEO/AEO structuring). White-label contract has no blocking terms.
Options considered
- Keep white-label and push them on quality: rejected, their incentive is throughput across many agencies, not per-client voice; the context library needed for non-generic content is structurally ours to build.
- Bolt a hosted CMS (WordPress etc.) onto client sites: rejected for JOLT! since the site is Next.js on Vercel; git-based MDX with PR-merge publishing is cheaper, versioned, and agent-friendly. Headless CMS (Sanity/Payload) remains an option for clients needing a UI.
Key design choices
- Pricing reframed around outcomes (“up to X articles/mo, allocation driven by research”), not article-count promises.
- Gradual topic clusters (3-5 articles/week in 1-2 clusters, double down on winners), not volume blasts.
- Every topic tagged GEO-first or SEO-first at research time.
- Monthly press releases replaced by monthly linkable assets (data studies, reports) with client input via their dedicated Slack channels (client-facing questions NOT enabled yet).
- Drafter and reviewer are separate model contexts; max 2 AI review rounds then human. Human reviewer: Ben (U07561UQ90S) in lore-content-review.
- Human edit diffs captured as exemplars (voice training data).
- YouTube: embed found videos now; produced videos later as a paid add-on tier.
Consequences
- Pipeline workspace
~/lore-content/on the Mini (architecture.md = decision record). - Phase 0: JOLT! end to end. Phase 1: 1-2 friendly clients in parallel with white-label. Phase 2: migrate the book, YouTube add-on, pre-sales audit generator.
- Margin impact once migrated: COGS drops from ~50% of revenue to infra + review time. Scoreboard note: after Phase 1, propose adding Lore gross margin to the per-entity strip.