Last updated April 18, 2024
What is a Transparency Note?
An AI system encompasses not just the technology itself, but also its users, those impacted by it, and the context in which it operates. Building a system that aligns with its intended purpose requires comprehending its inner workings, understanding its capabilities and constraints, and striving for optimal performance. SupportLogic’s Transparency Notes serve to illuminate how our AI technology functions, the decisions system owners can make that impact its performance and behavior, and the significance of considering the entire ecosystem, including technology, individuals, and the operational environment. You can leverage Transparency Notes during the development or deployment of your system, or share them with stakeholders who will interact with or be influenced by your system.
SupportLogic’s Transparency Notes are integral to our commitment to actualize our AI Principles.
Introduction
SupportLogic uses a mixture of Private and Public models. Our proprietary AI services (algorithms and models) are purpose-built and owned by SupportLogic. We only make use of OpenAI/Anthropic business or enterprise products for Gen AI functionality for Case Summarization, Case Resolution Detection, and the Gen AI functionality within Agent SX.
| CoreSX | ElevateSX | AgentSX | |
| Case Summarization | X | X | |
| Case Resolution | X | X |
These are the only product-features that are making use of Gen AI functionality. Opting-out of this functionality opts-out of the use of GenAI in your SupportLogic instance.
How is that data used by Third-Party GenAI providers?
Both OpenAI and Anthropic Business/Enterprise APIs do not use business data, inputs, or outputs for training models. As we use enterprise versions of both these services which are completely stateless and do not store any data or use that data to train ML models.
How is the data used to Train our proprietary AI models?
The data used by SupportLogic to train ML models is primarily composed of case metadata and/or data created within the SupportLogic data platform – we do not use any native CRM data to train machine learning models across environments.
We do not use PII data for training ML models; we use de-identified case metadata (case ID, account id, user id etc) to build these proprietary ML models.
How is data security/privacy preserved using SupportLogic AI models?
SupportLogic creates a dedicated VPC instance per customer and all ML models are hosted within the SupportLogic VPC environment.
- Our Intelligent Case Assignment, Account Health Scoring and Automatic Case Evaluation (for agent coaching) machine learning models reside within customer specific VPC instance
- Our signal extraction, scoring and escalation prediction engines run within SupportLogic VPC environment as a service
In addition, SupportLogic is also building self-hosted private GenAI/LLM models, and are planning to release this feature in the first half of 2024.