Start free. Upgrade when one campaign pays for the whole year. Every plan ships with the rewriter, scoring, lint, subject lines, and CTA grader — no nickel-and-dimed feature tiers.

Full AI generations count as 1 run. Lightweight passes (score, lint, CTA, preheader) count as 0.25.
| Feature | Free | Starter | Pro | Team |
|---|---|---|---|---|
| Monthly Sonnet runs | 10 | 250 | 1,500 | 5,000 pooled |
| Rewrite + 3 variants | ✓ | ✓ | ✓ | ✓ |
| Score + lint | ✓ | ✓ | ✓ | ✓ |
| Subject / preheader / CTA | ✓ | ✓ | ✓ | ✓ |
| Templates | — | ✓ | ✓ | Shared |
| Brand voice | — | 1 | 5 | Org-wide |
| Personas | — | 5 | Unlimited | Unlimited |
| Sequences | — | — | ✓ | ✓ |
| Roast / teardown / A/B predictor | — | — | ✓ | ✓ |
| Localization (ES/FR/DE/PT-BR) | — | — | ✓ | ✓ |
| Version history | — | — | ✓ | ✓ |
| Seats | 1 | 1 | 1 | 5 |
| API + webhooks | — | — | — | ✓ |
| 2FA | — | — | ✓ | ✓ |
| Email support | Community | 48h | 24h | Priority |
No. We are intentionally not an ESP. We optimize your copy; you keep sending where you already send (Mailchimp, HubSpot, Resend, your CRM, anywhere).
Hybrid: instant local lints (spam words, readability, link audit) plus a Haiku pass for clarity, CTA strength, and persuasion structure. Composite is weighted by email kind.
No. We never train on customer content. Drafts live in your workspace until you delete them.
It's a calibrated rubric, not an ML model trained on engagement data. We label it ‘predicted’ for that reason. Good for relative comparison, not absolute promises.
The product is built on Dokploy-friendly containers (external Postgres, Redis, MinIO). Enterprise self-host is on the roadmap.