Brand‑voice AI on Anthropic Sonnet 4.6 with a one‑hour prompt cache. A calendar that knows the difference between a draft and a live post. Federated platforms on day one, the rest as App Reviews clear. Built for marketing teams that have outgrown Buffer but refuse to pay Sprout's invoice.
Most schedulers treat content as a pile of cards to ship. contentOS starts where editorial teams have always started — with pillars, angles, and an assignable beat. The calendar inherits the strategy. Nothing gets queued without a topic and a why.
Anthropic Claude Sonnet 4.6 with a 1‑hour ephemeral cache on your brand voice template. After the first generation each hour, every variation reuses the same prefix at one tenth of the cost. We surface the model name, the cache state, and the per-piece cost on every draft. No sparkles.
Drag and drop across month / week / list views. The status of each piece is legible at a glance — Draft, In Review, Approved, Scheduled, Published — with a categorized state machine underneath. Approvals are per-piece, not per-role, so the right reviewer sees the right card.
Then Instagram, LinkedIn, TikTok, Pinterest as App Reviews clear. X is bring‑your‑own‑key — we never absorb pay-per-use. Every publisher uses the same idempotency key, the same retry curve, the same dead-letter queue. If the platform goes down at 9 a.m. on a Tuesday, your post recovers; you don't.
A separate worker container with pg-boss for scheduled publishes. SKIP LOCKED claims, two-phase publish with audit log, partial-publish reaper every 5 minutes. The web container can redeploy mid-issue without dropping a single in-flight job. The stuff you only notice when it's missing.
Four tiers. The Pro tier is the one we built the product for — mid-market teams displaced by the Jasper collapse and the Sprout invoice. The other three exist so the entry doesn't slam in your face.
For testing the voice. One channel, federated only.
Solo creators and freelancers running their own beat.
Mid-market marketing teams. The hole in the market.