Spark AI in Risk Cloud

Introduction to Spark AI

Spark AI is a set of capabilities within Risk Cloud that incorporates AI to streamline your processes, improve your decision making, and enhances how you do work in Risk Cloud. Spark AI is available to you and to your end users and and is completely optional and opt-in at every level.  

Need help getting internal approval to enable Spark AI? Visit Convince Your Boss: Why Spark AI Is Safe, Secure, and Ready for Governance Approval for AI governance-focused materials and success stories.

Enabling Spark AI

To be able to use the AI Integration it first needs to be enabled in your environment from the Integrations page. Risk Cloud Admin Users can access the Integrations page by navigating to "Admin" > "Integrations".

Once you enable Spark AI, you can then click "Configure" to enable individual Spark AI capabilities. 

Spark AI is powered by OpenAI's advanced models.

Spark AI Features

The following articles have more details on specific Spark AI capabilities.

• Text Assistant
• Record Linking Recommendations
• Autofill 
• Automated Evidence Testing (coming soon)
• Reporting Insights (coming soon)

Frequently Asked Questions

1. How does Spark AI use customer data?

Spark AI processes specified fields (e.g., text, rich text, and concatenated fields) to generate recommendations, using either LogicGate’s OpenAI API or a customer-provided OpenAI API. For LogicGate’s API, API request and response data is retained by OpenAI for 30 days as outlined in OpenAI's platform documents but is not used to train models. Spark AI only uses customer data within the customer’s instance and does not share or use it to serve other customers. The use of your own API keys would allow you to have direct authority over your data with OpenAI.

2. Which Large Language Models (LLMs) are used?

Spark AI features use a combination of OpenAI GPT and embedding models, selected to deliver the best balance of accuracy and performance for each use case. We regularly evaluate and update the models in use to ensure the best fit as capabilities evolve. The following OpenAI models currently support Spark AI features:

For the most up-to-date model versions, refer to Integrations > Spark AI  > Configure in Risk Cloud.

3. Can I disable Spark AI if my organization is not ready to use it?

Yes, Spark AI features are entirely optional and opt-in, we ensure your organization always know when and where AI is in use. Administrators (and only administrators) can enable or disable Spark AI features within Risk Cloud via the “Admin” > “Integrations” menu at any time.

4. What is the relationship between LogicGate and OpenAI?

The relationship between LogicGate and OpenAI is governed by OpenAI’s Business Terms, which outline the rights and responsibilities associated with using OpenAI’s services. These terms apply to LogicGate’s integration of OpenAI’s models into the Risk Cloud platform, enabling features like Spark AI. Key aspects include:

• Use of Services: OpenAI grants LogicGate a non-exclusive right to access and use its services, including APIs, to integrate into applications like Risk Cloud.
• Content: LogicGate may provide input to OpenAI’s services and receive output, collectively termed “Customer Content.” OpenAI’s Business Terms specify how this content is handled and the rights associated with it. It also details how they will process and store Customer Content in accordance with their Enterprise privacy⁠ commitments.
• Data Usage and Retention: OpenAI’s Business Terms detail data usage policies, including retention periods and purposes for which data may be used. By default, OpenAI retains data for 30 days and does not use it to train its models. More detail can be found in the Enterprise privacy⁠ commitments and platform documentation.

Spark AI features also allows customers to optionally use their own OpenAI API key, in which case you would directly have a relationship with OpenAI.

5. How is Spark AI configured and secured?

The following measurements and configurations are implemented to safeguard your data and Spark AI usage in Risk Cloud.

• Enabling Spark AI: Administrators (and only administrators) can activate Spark AI within Risk Cloud by navigating to “Admin” > “Integrations” and enabling the AI integration at each feature level (e.g., AI Text Assistant, Record Linking Recommendations).
• Data Processing: Spark AI processes user text-based inputs for the following purposes: data analysis, automated reporting, generating insights and recommendations, and other purposes relevant to the AI Services.. LogicGate logs these inputs to monitor for inappropriate use and to enhance product functionality. Importantly, LogicGate does not use your input or output for any other purpose such as training a model.(LogicGate Subprocessors & Affiliates)
• Data Retention: By default, LogicGate’s OpenAI key uses OpenAI’s standard data retention policy of 30 days. During this period, data is not used to train OpenAI’s models. For the AI Text Assistant, OpenAI only receives data provided via prompts. For Record Linking Recommendations, OpenAI processes data from your records (currently only limited to text, text areas, and concatenated fields), which you can review and approve when configuring Linked Workflow Sections. (LogicGate Help)
• API Key Management: By default, the Spark AI features utilize LogicGate’s OpenAI API key. However, you have the option to use your own OpenAI API key, granting you direct authority over your data with OpenAI. (LogicGate Help)
• Information Security & Privacy: You can find more information about how LogicGate protects your data at our Trust Center. 

For more questions or information, please contact your account team contact.