Workspace access and compute isolation
Client access is scoped to its workspace. Computational tasks run in isolated sandboxes associated with that workspace, helping reduce unintended exposure between research environments.
Protection across research workflows
Self-hosted model data use
Our research model is self-hosted and grounded for M3TRIQ workflows. It does not retain research inputs or use them for training. When a workflow needs additional isolation, model inference runs inside an isolated compute sandbox.
Operational diagnostics
We retain issue and error logs to investigate failures and improve reliability. Routine troubleshooting uses these diagnostics and does not involve accessing the underlying research files or intellectual property.
Research data and platform storage
The model’s input-handling policy is separate from storage by the platform. Some M3TRIQ features save project files, datasets, session content, and results in a workspace so teams can return to their work. This means that “not retained by the model” does not mean that every platform feature deletes all research content after each interaction.
Privacy and AI-security guidance recommends limiting data use to defined purposes and considering disclosure risk across the full application, not just during model training. For a workflow review, research teams should consider what data is needed, where it is processed, what is saved, who can access it, and how long it is retained. Contact us to discuss a specific use case or retention question.
These safeguards are designed to reduce exposure; no online service can guarantee zero security risk.
Questions about data handling?
Contact our team to discuss how M3TRIQ handles research data in your workflow.
Contact M3TRIQ