AI Usage and Transparency Policy

Our AI Coach functions through isolated, enterprise-grade APIs provisioned by third-party foundational model providers (such as OpenAI, Anthropic, or Google).

The Zero-Training Mandate: All user prompt texts and scenario responses transmitted to the AI engines are encrypted in transit. Our enterprise vendor agreements legally guarantee that these AI providers cannot and will not use Crescaa user data, inputs, or generated outputs to train, refine, or optimize their public machine-learning models. Your data is processed transiently to generate a response and is immediately discarded by the vendor.

Automated Redaction & Prompt Engineering: As a secondary technical safeguard to protect third-party privacy, Crescaa utilizes system-level prompt engineering designed to detect and ignore suspected real names, direct contact information, or specific geographic locations before your notes are processed by the AI APIs.

Despite our automated technical safeguards, you are required to maintain privacy-by-design at the user interface level. You must pseudonymize all third-party subjects before querying the AI Coach or logging notes. You are strictly required to use the designated "Teammate" alias/nickname format (e.g., "Agent A"). Do not feed real legal names, email addresses, or identifiable HR records into the AI prompts.

Under the EU AI Act, AI tools utilized within human resource or managerial contexts carry distinct compliance obligations. Crescaa mitigates these risks by operating strictly as an educational advisory tool, not a formal HR Information System (HRIS).

No Automated Decisions: The system is structurally and legally prohibited from executing automated decision-making processes. The platform cannot unilaterally issue employment reprimands, demote staff, execute payroll changes, or make hiring/firing determinations.

The Independent Review Requirement: To combat "automation bias," all behavioral profiles, conflict resolutions, or management action plans suggested by the AI Coach are strictly draft educational materials. You have an absolute legal and organizational obligation to critically analyze, review, alter, and validate every AI recommendation using independent human judgment before executing any action in your workplace.

Generative artificial intelligence predicts token sequences; it does not comprehend contextual truth or factual reality. By using the AI elements of the Service, you acknowledge:

Hallucination Risk: The AI Coach can generate text that appears highly authoritative but is factually inaccurate, legally non-compliant, or fundamentally wrong for your specific situation.

Lack of Organizational Context: The AI Coach cannot understand the unique local labor laws, trade union bargaining contracts, cultural nuances, or internal corporate policies of your specific workplace. You are solely responsible for ensuring that your application of any AI advice complies with your local jurisdiction.

You are strictly prohibited from utilizing the AI Coach to:

Generate psychological assessments designed to manipulate, exploit, stalk, or retaliate against colleagues or direct reports.

Optimize prompts to discriminate against team members based on protected demographic criteria (including race, gender identity, age, physical disability, religion, or sexual orientation).

Fabricate falsified disciplinary records, generate automated performance reviews without objective human tracking data, or draft formal Performance Improvement Plans (PIPs) meant for official HR submission.

Execute prompt injection attacks, jailbreaks, or attempt to reverse-engineer our system instructions, logic constraints, or API keys.