Encrypted inferencing with Cognita

[ac-aisec]

Hi,

We would like to enhance Cognita RAG use case by integrating MirrorSecurity SDK.

The MirrorSecurity SDK provides the following key features that we believe will significantly benefit Cognita:

VectaX: This feature delivers enterprise-grade security for vector operations through similarity-preserving encryption. It ensures that embeddings remain protected while still allowing for efficient searchability—a critical capability for maintaining both security and performance in our application.

AgentIQ: This component bolsters generative AI security by incorporating essential features such as guardrails, threat detection, compliance enforcement, and policy management.

For context, we have already successfully integrated Mirror VectaX with the Qdrant database. You can find detailed information about this integration in this blog post, which outlines how MirrorSecurity collaborates with Qdrant to enhance functionality.

We would greatly appreciate your insights on potential approaches for seamlessly integrating MirrorSecurity into Cognita.

Thanks, Anand