Skip to main content

​
Underlying Models

We work with our own models as needed or third-party foundation models, configured so customer data is not used for training.

​
Access Controls

Role-based access, least-privilege internally, and no standing access to customer data.

​
Audit Logging

Every action is logged: what happened, when, and what triggered it.

​
Incident Response

We notify affected customers within 24 hours of a confirmed incident.

​
Compliance

  • SOC 2 Type II certified
  • GDPR compliant
  • CCPA compliant
  • Enterprise-grade security

​
Evaluation & Feedback

We evaluate AI output quality before deployment and monitor it continuously in production, testing draft accuracy, relevance, and adherence to configured bounds against real-world scenarios. Customers can flag any AI output directly in the product or through their Emanate account team. Flagged outputs are reviewed by our engineering team, and confirmed issues feed directly into model configuration, prompt, and guardrail improvements. Changes to underlying models or system behavior are tested against our evaluation suite before release.