Exabeam Nova enhances security operations with AI-powered capabilities by leveraging Google Gemini Large Language Models (LLMs) within Google Cloud Platform (GCP), where our New-Scale Analytics also operates. When Nova is activated, it securely retrieves relevant data—such as threat detections, timelines, and related entities—from the New-Scale platform.
This data is structured into a comprehensive prompt and encrypted using standard SSL protocols before being sent to a pre-trained Gemini model for processing. To maintain compliance and optimize performance, model processing is performed within the customer’s designated geographic region whenever possible.
Crucially, Exabeam employs pre-trained Gemini models, meaning customer data is never used for model training. Additionally, the entire evaluation process occurs in memory and is fully ephemeral—data is never stored, cached, or retained after processing. Once the model generates a response, it is encrypted and immediately returned to New-Scale Analytics, ensuring no persistent data footprint within Gemini.
Examples of Nova in action include investigation and advisory summaries as well as the analyst assistant, both of which are available in Threat Center and Outcomes Navigator.