1. Our "No-Training" Commitment

We believe your data belongs to you.

• Zero Model Training: The Pennar LLC does not use client interaction data, transcripts, or user-provided media to train our proprietary rendering models or third-party LLMs.
• Transient Processing: For our international Maya API, user audio and video are processed in real-time memory and purged immediately following the session.

2. Data Residency & Sovereignty

We align our infrastructure with the geographic requirements of our partners:

• Aria (US/Local): Data associated with our local deployments is stored and processed exclusively on US-based servers, utilizing Google Cloud’s restricted regions for government-grade compliance.
• Maya (International): We utilize Cloudflare’s Regional Services to ensure data stays within its origin jurisdiction (e.g., GDPR-compliant nodes for EU users), satisfying local data residency laws.

3. Technical Safeguards

• Encryption Everywhere: All data is encrypted using AES-256 at rest and TLS 1.3 in transit.
• PII Redaction: Our orchestration layer automatically masks Personally Identifiable Information (PII) before it reaches the reasoning engine (Google Gemini), ensuring that sensitive user details are never exposed to the LLM.
• Edge Security: By leveraging Cloudflare’s WAF (Web Application Firewall), we provide specialized protection against SQL injection, DDoS attacks, and prompt-injection attempts.

4. AI Ethics & Safety

Our integration with Google Gemini as the common intelligence core for all products allows us to leverage enterprise-level safety filters.

Aria and Maya are programmed with strict guardrails to prevent the generation of harmful, biased, or off-topic content, ensuring a professional representation of your brand or municipality at all times.