Built for $19. Scales to $99.
Average ARPU: $35.25/mo. 89% gross margin. No seat tax. Per-tenant LLM cost ceilings keep you safe even on a worst-day run-away prompt.
Starter
For new creators publishing 1-2 pieces a week.
$182 billed annually
- 8 AI long-form articles / month
- Module 1 · Topic Discovery (basic)
- Module 3 · Draft Engine
- Module 4 · SEO score (live)
- Cross-platform repurpose: X + LinkedIn
- 1 brand voice profile
- 1 connected publication
- Voice match auto-rewrite loop
- Google Search Console
- AppSumo lifetime
Pro
Where most full-time creators land. Unlocks the full agent pipeline.
$470 billed annually
- 25 AI long-form articles / month
- Everything in Starter, plus:
- Module 2 · SERP-Driven Outline (full)
- Voice Lock auto-rewrite loop
- All 7 cross-platform repurpose adapters
- Module 6 · Subscriber Conversion Toolkit
- 3 brand voice profiles
- 5 connected publications
Business
For multi-brand content studios and serious SEO operators.
$950 billed annually
- 80 AI long-form articles / month
- Everything in Pro, plus:
- Google Search Console keyword tracking
- Programmatic SEO bulk generator
- Unlimited brand voice profiles & projects
- Priority Sonnet 4.5 routing
- API access (read + draft)
- Priority email + community support
Lemon Squeezy is our Merchant of Record — they handle EU VAT, US sales tax, and refunds. Stack up to 3 AppSumo lifetime codes for a permanent Pro / Business unlock.
Tell me how this is actually different.
Is the output really not 'AI-flavoured'?+
We train a per-user Voice Profile from 3 of your past articles, embed it with text-embedding-3-small, extract a style rubric (sentence length, hedging, POV, signature phrases), and gate every generated section behind a Voice-Match score. Sections under threshold are auto-rewritten with the rubric in the prompt — not just temperature tweaked.
Which newsletter platforms do you publish to?+
Beehiiv (API), Ghost (Admin API), Kit / ConvertKit (API). Substack lacks a public publish API, so we provide a one-click rich-text copy with metadata pre-filled. Every integration is OAuth-based with encrypted-at-rest tokens; you can revoke at any time from /integrations.
How is the 7-step agent different from chaining ChatGPT prompts?+
Three differences. (1) Durability: the pipeline runs on Inngest, so a failed step retries from the last good checkpoint instead of losing 6 minutes of work. (2) Cost-aware routing: each step picks the cheapest model that hits its rubric; expensive Sonnet calls only happen for Voice-Match rewrites and final polish on Pro/Business tiers. (3) Auditability: every intermediate output is editable, persisted, and re-runnable — you never lose visibility.
What's the cost ceiling? Can a runaway prompt blow up my bill?+
Per-tenant monthly USD ceilings: $4 Starter / $18 Pro / $70 Business — enforced in the LLM router itself. At 80% of the ceiling, the router hard-falls-back to gpt-4o-mini + DeepSeek even on Business, and posts an alert to your /admin (and Sentry). Quotas are also article-based: 8 / 25 / 80 long-form pieces per month.
Do you train on my content?+
No. We don't fine-tune any model on user content. Embeddings of your Voice Profile are stored in your own row in pgvector, used only at inference time for your account, and deleted on account close. We sign Data Processing Addenda for Business customers.
Refund policy?+
30-day no-questions-asked refund on monthly plans. AppSumo lifetime codes follow AppSumo's 60-day refund policy. Annual plans pro-rate.
Why GitHub + Vercel + Neon?+
Because solo + bootstrap. Vercel + Neon is the lowest-friction, highest-quality serverless Postgres + Edge stack in 2026. We never want to spend a Saturday on cluster maintenance.
What happens if I cancel?+
You keep edit access to existing drafts, can export everything (Markdown, HTML, JSON) for 90 days via /settings, and your data is purged on day-91 unless you opt to extend retention.