Spark AI
Go fast and safe with our innovative suite of opt-in features designed to bring intelligence, speed, and precision to your GRC processes. With Spark AI, you’re not just keeping up with the complexities of modern risk management, you’re mastering them.
Add a Little Spark to Risk Cloud®
Spark AI is a suite of AI-powered, opt-in features designed to enhance the experience of program owners and end‑users across their interactions with Risk Cloud.
Key Capabilities and Product Features
Leverage the power of generative AI to automatically draft risk and compliance materials that intelligently reference related program data, GRC best practices, and consistent writing standards.
Say good-bye to manual gap analyses and bost control cross-mapping accuracy so you can scale and mature your compliance programs with ease.
Unify efforts and harness the full potential of your GRC data by connecting risk and compliance insights across the entire organization.
Frequently Asked Questions
Released: Record Linking Recommendations for End Users (Beta)
End users will receive recommendations for all records linked across all workflows in their Risk Cloud environment. Customers must opt-in to use Spark AI Record Linking Recommendations. Contact your account team for more information.
Later: Record Linking Recommendations for Builders
Builders will have more granular configurations to control specific workflows to receive linking recommendations.
Yes, Spark AI features, including Text Assistant and RLR will be included in customers’ current subscription without additional costs.
By opting into Spark AI, OpenAI processes specific fields—such as text, text areas, and concatenated fields from linked records—to generate recommendations. You can use either your own OpenAI API or LogicGate’s OpenAI API. If your organization has a separate agreement with OpenAI (e.g., zero-day retention), that policy will apply.
When using LogicGate’s OpenAI API, OpenAI follows its standard 30-day data retention policy, deleting data after 30 days. Data sent via LogicGate’s OpenAI API is not used to train or improve OpenAI’s models. We are also working on additional product enhancements to let builders control which fields are used for Spark AI RLR.
Match strength is calculated using the cosine similarity score between the vector embeddings of the parent record (e.g., your internal control) and candidate child records (e.g., framework controls). Vector embeddings capture the semantic meaning of the records. A match strength above 50% currently indicates a strong match. The team is actively exploring several approaches to further refine the recommendation methodology.
Yes, we leverage RAG (Retrieval Augmented Generation) to reference the parent record (e.g., your internal control) and its meaning. This helps generate more consistent recommendations for mapping child records (e.g., framework controls).
Resources
As artificial intelligence (AI) continues to advance rapidly and organizations expand their usage to optimize efficiency and productivity, implementing internal AI policies to ensure regulatory compliance and minimize exposure remains a hot topic.