Sovereign hosting, and no shared models.
Your process data is the asset. It stays on our own infrastructure, it never trains a model anyone else uses, and it leaves permanently when you ask.
- Hosting
- Our own servers in France and the United States, with no third-party cloud subprocessor between you and your data.
- Model isolation
- No client data is used to train models shared with other clients. What your plant produces improves your plant’s models and nobody else’s.
- Deletion
- Permanent deletion of your data and of every model derived from it, on request.
- Access control
- SAML SSO and role-based access, so the plant decides who sees which line.
- Certification
- ISO 27001 and SOC 2 are in their final stages. We will keep saying exactly that until they are signed.
For anomaly detection, micro-stop counts only.
The aerospace and defence deployment ran in a regulated environment on the run/stop event log alone.
What we read
How many micro-stops, how long they lasted, and whether they came in bursts or evenly spread. Counts and durations.
What we do not read
No named operator data. Nothing that identifies who was on the line when a stop happened.
What that means for you
Your GDPR scope is unchanged by the deployment. There is no new category of personal data to declare.
Two properties that are not configuration.
Both are consequences of how the models are built, so they cannot drift out of compliance later.
Only the past is ever used
Every score is produced from information available at that moment and no later. That is what makes offline validation meaningful, and it also means a score can never be explained by data the line did not yet have.
Nothing runs outside our infrastructure
Scoring is one lightweight computation per machine every ten minutes. There is no GPU and no call to an external service, so there is no third party in the data path to audit.
How results are tested before they reach you is on How we validate.
Send the questionnaire. We answer in writing.
If your team has a security review process, start it at the first conversation rather than after. Tell us what you need and we will work through it.
