Skip to main content
All data is stored in the EU (eu-west-2) on Supabase-managed PostgreSQL with row-level security enabled. Infrastructure runs on AWS, Vercel, and Fly.io, all EU-hosted.
Your seed data is used to build and operate your user model. That is what the service does — we train a model for you, from your data, scoped to your account.Your data is never used to:
  • train, fine-tune or otherwise adjust any other customer’s user model;
  • make your user model accessible to any other customer (each user model is scoped to its owning account);
  • train foundation or general-purpose models.
Third-party model providers we call at prediction time do not retain your input questions or user model data.Separately, we do use aggregate performance data about user models (accuracy scores, test results, evaluation metrics) to improve the methodology by which we build user models in general. That is not training on your content — it is the same kind of learning a research team does when reviewing which methods worked. The aggregate statistics we keep do not contain your data and cannot be reversed back to it.
API keys are generated per account and scoped to your organisation. Keys can be rotated at any time from the dashboard. All API requests are authenticated and rate-limited.
We run a mix of our own inference infrastructure and major model providers with enterprise data processing agreements. Your input questions and audience data are sent to the model only at prediction time and are not retained by any provider.
Not yet. We’re a small team building toward formal certification as we scale. If you have specific compliance requirements, get in touch and we’ll work through them with you.
Yes. You can delete user models, predictions, and your account at any time. Deletion is permanent and propagates to all storage layers.