AI Usage & Responsible AI Policy

Last Updated: October 12, 2026 · Entity: YM PROPERTYPATH - SMC LTD

This AI Usage / Responsible AI Policy describes how YM PROPERTYPATH - SMC LTD designs, deploys, and governs AI features (including Maya and customer-trained agents) across the Patriy platform.

1. Our Principles

• Helpful & accurate: Prefer grounded answers from your knowledge base and CRM context over invention.
• Transparent: Support disclosure that users are interacting with AI where required or expected.
• Safe: Block abuse, harmful content, and policy violations via automated and human-assisted controls.
• Controllable: You own prompts, knowledge, channel configs, handoff rules, and escalation to humans.
• Privacy-aware: Minimize unnecessary personal data in model context; honor deletion and access rights.

2. What Patriy AI Does

• Answers customer questions across WhatsApp, Instagram, Email, Telegram, X, and other connected channels (WhatsApp: new connections paused; Instagram & Messenger: in development — live channel status)
• Qualifies leads, books meetings, and logs interactions into CRM
• Assists marketing/content workflows (e.g. X Growth Studio) under your configuration
• Provides an interactive product demo (Maya) with live or fallback responses

3. What Patriy AI Is Not

• Not a substitute for licensed professional advice (legal, medical, financial) unless you explicitly configure domain content and assume responsibility
• Not guaranteed error-free; outputs can be incomplete or incorrect
• Not autonomous decision-making for employment, credit, or other high-risk automated decisions without your human review processes

4. Customer Responsibilities

• Upload accurate FAQs, policies, and brand voice
• Set handoff rules for sensitive or complex issues
• Comply with channel rules and industry regulations
• Obtain required consents for messaging and data use
• Review analytics and sample transcripts for quality

5. Safety & Guardrails

• Policy guardian / content filters for disallowed content categories
• Rate limits and anti-abuse controls
• Human takeover from the unified inbox
• Enterprise options for stricter data handling and access control

6. Training Data & Model Improvement

We use xAI (Grok) as our AI model provider under contractual terms. Customer conversation content is processed to provide the Service. We do not sell your customer message content. Any use of aggregated or de-identified insights for product improvement will respect our Privacy Policy and DPA. Enterprise customers may negotiate stricter data-use terms.

7. Bias, Fairness & Accessibility

We work to reduce harmful bias and support multi-language service. You should monitor agent behavior in your markets and correct knowledge-base gaps that cause unfair or inaccurate responses.

8. Incident Response

Report harmful AI behavior to hello@patriy.com. We investigate and may disable misconfigured agents or abusive accounts.

9. Related Documents

• Acceptable Use Policy
• Privacy Policy
• Data Processing Agreement
• Security Center