What it is
AI Review is a personalisation and verification layer that sits between AI-generated content and its audience. You bring AI output from any tool (ChatGPT, Claude, an internal agent), pass it through AI Review, and get back the same content rewritten in your voice — with AI-tells stripped, hard rules enforced, and a transparency report on what changed.
Why we built it
Knowledge workers use multiple LLMs — office-provided, personal, agent-driven. Each tool produces output in a generic AI voice. People manually rewrite to sound like themselves. AI Review fixes that with a per-persona voice model and a stable API you can drop into any workflow.
Status: open beta
We're in open beta. Sign-up is free. Bring your own provider key, or use the shared free-trial budget (50 requests/day, 20k tokens/day). The following are deferred to v1.1:
- Email verification + password reset (no email transport in v1)
- OAuth (Google / Microsoft)
- Multi-member orgs with invites
- Stripe billing
- Persona "template" library (Work / Personal / LinkedIn presets)
- Reference / fact verifier
- Custom OpenAI-compatible endpoint (Ollama / vLLM self-host)
Tech
- Next.js 15 (App Router) on Vercel
- MongoDB Atlas (M0 free tier)
- NextAuth v5 (credentials)
- DeepSeek / Anthropic / OpenAI SDKs
- Upstash Redis for rate limiting
- OpenAPI 3.0 spec at
/api/openapi.json, Scalar docs at/docs