Overview

At Perceptyx, we are committed to responsible AI practices that prioritize data privacy and security. We do not use customer-specific data in our AI model training, and we employ anonymization and security protocols designed to protect customer data while delivering powerful, ethical AI-driven insights.

On-demand coaching is an AI-powered development tool designed to support leadership growth while maintaining security, transparency, and responsible AI governance. Below is a high-level overview of our approach to data protection, AI governance, and security to help HR leaders and IT teams evaluate how on-demand coaching might fit into their company’s strategy.

Data Privacy & Security

1. Enterprise-Grade Protection: On-demand coaching encrypts data in transit and at rest. Perceptyx’s current certifications include ISO/IEC 27701, ISO/IEC 27001, SOC 2 Type 2, FedRAMP, and Data Privacy Framework.

2. Data Usage Transparency: To provide AI-powered coaching, on-demand coaching passes user inputs to OpenAI, a Perceptyx subprocessor, via a secure API. Under OpenAI’s business terms, OpenAI does not use any customer data for training their AI models. Perceptyx also does not use customer data for training our AI models. Perceptyx uses anonymized interaction data only for product development purposes to improve the user experience. On-demand coaching uses and keeps data already stored within the Perceptyx platform, such as HRIS data and action planning data, to personalize its interactions with each employee.  

3. Data Privacy: Perceptyx does not collect or transmit any personally identifiable information (PII) or individual-level survey response data to provide on-demand coaching. On-demand coaching is an “open chat” interface, so any information that users choose to include in their messages to on-demand coaching will be sent to OpenAI; this could potentially include PII. 

4. Data anonymization and minimal retention: OpenAI securely retains API inputs and outputs for up to 30 days as part of their own standard logging process; after that, data is stored only within internal Perceptyx systems. Conversation data in Perceptyx systems is only accessible to select Perceptyx employees with responsibilities for maintaining and improving the on-demand coaching experience. All user data is obfuscated for anonymization. Perceptyx will never share individual-level conversation data with customers; only anonymized, aggregated data will be available for analysis by customers. 

5. Authentication & Role-Based Access: Supports SSO, multi-factor authentication (MFA), and role-based permissions to limit access. On-demand coaching access can be configured at the user level, ensuring only authorized employees can use it. Access to on-demand coaching can be configured separately from access to Intelligent Nudges and other Perceptyx features. 

6. Continuous Monitoring: Security is reinforced through logging, threat detection, and real-time monitoring.

Responsible AI & HR-Specific Guardrails

1. Designed for Development: On-demand coaching is strictly for employee growth and development, not performance evaluations or disciplinary actions.

2. Trained on Curated Behavioral Science Research: On-demand coaching provides trustworthy guidance based on Perceptyx’s library of curated research and science-backed nudges.   

3. Guardrails for Safe Use: On-demand coaching will refuse to answer irrelevant or inappropriate questions and will direct employees to HR for sensitive issues like harassment concerns.

4. Transparency & Explainability: On-demand coaching provides a transparent record of how recommendations are generated, consistent with ethical AI principles. 

On-demand coaching prioritizes security, privacy, and ethical AI, so HR leaders can confidently deploy AI-powered coaching while protecting employee trust and data integrity. For further details or a security review, please reach out to our team.