Skip to main content

​
Does Semilattice process personal data?

Only to run your account: the personal data we store is your account information, listed below. Planning and running simulations needs no personal data, and the data user models are built from should not contain any.

​
Are simulated users real people?

No. Each simulated user is a model trained on what one real person said and did in the data a user model is built from, and it predicts how that person would answer a question they were never asked. It is not that person, and nothing it answers identifies them. No real person is contacted or observed when a simulation runs.

​
What data do you collect about me as a user?

Account information (email, organisation), which pages and features you use, recorded by our own servers, and error reports. We don’t sell user data or share it with third parties for marketing purposes.

​
How is the content I upload for models different from the usage data you collect?

Customer Model Data is the content you provide to build a user model of your own users: interviews, survey responses, reviews and other files you send us for that purpose. It is used only to build and operate your user models and to run your simulations. It is never used to train other customers’ models, never made available to other customers, and never used to train foundation models. It is governed by our Terms of Service. Customer Model Data should not contain personal data; if any personal data is erroneously included, our Data Processing Agreement applies to that personal data. Service Usage Data is how you use our apps — login events, feature usage, performance metrics, error reports, product analytics. We use it to operate, secure, improve and develop the service. It is governed by our Privacy Policy.

​
Is Semilattice GDPR compliant?

We are a UK-based company processing data in the EU. We follow GDPR principles: data minimisation, purpose limitation, and the right to deletion. Our infrastructure is EU-hosted. See Security for where data is stored. For a formal data processing agreement, contact us.

​
Can a simulated user’s answers be traced back to a real person?

No. You can read how each simulated user answered, but those answers are predictions, not records of what anyone said, and there is no way to work back from them to the real person a simulated user is modelled on. User models are built from data without names, emails or other details that identify someone.