This AI Policy describes how Cyrenza, Inc. (“Cyrenza”) develops, deploys, and governs artificial intelligence within the Cyrenza platform. This policy is incorporated into the Platform Agreement.
1.1. Cyrenza provides an AI Workforce Platform with domain-specialized Agents that perform professional knowledge work across multiple industry verticals. Our Agents are designed to augment human professionals – not replace them.
1.2. We are committed to building AI systems that are transparent, safe, and accountable.
2.1. Agent Operations. AI models process Customer inputs and data to generate work products (Artifacts) such as research, analysis, drafts, and reports.
2.2. Retrieval. AI-powered retrieval systems search and retrieve relevant information from Customer documents and data to provide context for Agent operations.
2.3. Context Compilation. AI assists in compiling and prioritizing relevant context from Customer Workspaces, documents, and prior interactions.
2.4. Content Safety. AI models are used to detect potentially harmful, biased, or non-compliant content in Agent outputs.
3.1. Cyrenza utilizes AI models from multiple providers to power its Agents. The specific model used for any given operation may vary based on task suitability, availability, and performance.
3.2. Current AI model providers are listed in our Subprocessor List.
3.3. Cyrenza may change the models or providers used for specific tasks without notice, provided the overall quality and capability of the Services are maintained.
4.1. Cyrenza does not use Customer Data to train, fine-tune, or improve its foundational AI models. Customer Data is processed solely to provide the Services.
4.2. AI model providers are contractually prohibited from using Customer Data for model training.
4.3. Customer Data sent to AI model providers for inference is processed transiently and is not retained by the provider beyond the duration of the request.
5.1. All Agent outputs are presented to users for review. Cyrenza does not automatically act on AI-generated outputs without human intervention.
5.2. Cyrenza does not make fully automated decisions that produce legal or similarly significant effects on individuals.
5.3. Customers are responsible for implementing appropriate human review processes commensurate with the risk and impact of decisions based on Agent outputs.
6.1. AI-generated outputs may contain errors, inaccuracies, omissions, or biased content. Outputs should be reviewed by qualified professionals before use in material business, legal, financial, or regulatory contexts.
6.2. Agent outputs do not constitute legal, financial, investment, tax, medical, insurance, or other professional advice. Cyrenza is not a licensed professional services provider.
6.3. The quality of outputs depends in part on the quality and completeness of Customer inputs and data.
7.1. Cyrenza provides provenance tracking for Agent outputs, including source attribution and lineage information, to support transparency and auditability.
7.2. Customers can review which sources and data were used to generate specific outputs.
8.1. Agent inputs and outputs pass through safety and content filtering layers designed to detect and mitigate harmful, biased, or non-compliant content.
8.2. Cyrenza implements compliance checks relevant to specific industry verticals, including protections against discriminatory content in regulated contexts.
8.3. Prompt injection protections and output validation are applied to guard against adversarial misuse of AI systems.
9.1. Cyrenza implements measures to detect and mitigate bias in AI outputs.
9.2. We monitor AI outputs for patterns of bias and take corrective action where issues are identified.
9.3. Customers are encouraged to report concerns about biased outputs to: ai-ethics@cyrenza.com
10.1. Customer Data sent to AI models is encrypted in transit.
10.2. Context compilation includes sensitivity controls that prevent unnecessary exposure of sensitive data to AI models.
10.3. All AI model interactions are logged for audit purposes.
10.4. Further details are provided in the Security Addendum.
Cyrenza may update this AI Policy from time to time. We will provide notice of material changes through the Platform or by email.