Trust
Trust.
Sovereign by design: we help customers train and run manifolds on their own infrastructure, and we never touch their data.
Sovereign by design
We help customers train a manifold on their own infrastructure, then license it to them to run there. We never touch the data: it stays under the customer's control, in the customer's jurisdiction, and the customer owns it. The customer is the data controller and publishes its own privacy terms. Where a customer wants more of its AI to run in-country, such as speech, voice or scoring models, we help build those models to run locally too.
What crosses the boundary
When we license a manifold, what ships is geometry: outcomes, prerequisite structure, progression and activity mappings. No records, trajectories, transcripts or audio travel with it, and nothing in it could identify or reconstruct a person.
Training stays local
A manifold learns where the evidence lives: inside the customer's own deployment. No personal data comes to BlueberryML to train or refine anything.
Outloud English
For Outloud, our own product, BlueberryML is the data controller. See Outloud's trust page.
Characters and companions
Our companions always say they're an AI. In care settings, contacts are confirmed by the home, content is vetted, and the home decides what is shared and with whom. Our companions aren't medical devices: they don't diagnose, treat, monitor or predict any condition.
This website
blueberryml.com uses privacy-friendly analytics (Vercel Analytics) and takes enquiries by email. See the Privacy Policy.
Questions
Data questions about BlueberryML: [email protected]. Questions about a deployment's data go to the client running it.