SCALT is an AI governance and control platform that sits between an organization's employees and the AI tools and models they use. It applies data loss prevention to stop confidential information from being sent to AI providers, detects and blocks prompt injection and misuse attempts, and filters inappropriate outputs.
Before deployment, each AI agent is evaluated and scored against test cases built from real company scenarios. Changes to agents go through review and approval, are rolled out progressively, and can be rolled back to a previous version. All interactions are logged and searchable for audit.
Policies are defined centrally and applied across all AI tools, regardless of the model provider. The platform also includes a no-code builder for creating and orchestrating AI agents connected to internal data sources.
SCALT is developed in France, hosted in Europe, and designed to help organizations meet GDPR and EU AI Act requirements.
