Instant Insights Data and Technologies for VRM and TMH

Instant Insights is a feature for the Vendor Risk Management and Trust Management Hub applications that uses generative AI to extract a SOC2 Type II report’s contents, analyzes them, and then summarizes the contents into deep and actionable insights.

• About the Data
• AI Technologies Used in Bitsight
• Data Security and Privacy

About the Data

The AI analyzes and extracts key information, such as:

• Type of report
• Nature of services provided
• Report period
• Locations where data is stored or processed
• Auditor’s name
• Auditor’s opinion (qualified/unqualified)
• Any flagged exceptions or risks

Each control includes references to the pages in the document where the information was sourced. You can also preview the document to confirm the accuracy of the AI responses.

AI Technologies Used in Bitsight

Instant Insights is powered by the Google Gemini model, integrated through Vertex AI. Leveraging the Gemini model's advanced capabilities, the analysis results provides:

• Deep insights: Extract valuable information and uncover hidden patterns within documents.
• Enhanced comprehension: Understand complex language and context within documents, leading to more precise analysis.
• Industry-leading security: Data is processed with the highest security standards, ensuring confidentiality and integrity.

Data Security and Privacy

We developed Instant Insights with security at the forefront to make sure document data is secure:

• Documents are securely handled and protected by Bitsight. Your sensitive information is kept confidential and secure throughout the analysis process.
• Data is not shared across vendors. Documents are analyzed independently; Data from one document or vendor is never used to influence the analysis of another. This ensures information from different vendors or reports remains isolated and are never intermingled.
• Third-party providers (such as Google) are strictly prohibited from using Bitsight customer information for training, fine-tuning, or improving any third-party AI models or services.