Privacy-First Inference
Built exclusively on Apple Silicon's Neural Engine, all AI operations occur on-device. Zero data, prompts, or telemetry are ever sent to external cloud servers.
- ✓ On-Device Generation: Leverages macOS 26+ Foundation Models capabilities.
- ✓ Intent Classification: Local CoreML intent router (`GuardianIntentClassifier.model`).
- ✓ Continual Learning: Low-Rank Adaptation (LoRA) subtly tunes the model to your environment.
- ✓ Isolated Processing: Completely separated from the active network shield to ensure 100% uptime.
Dynamic Personalities
Adaptive AI that speaks your language. PhantomGuardian features 5 distinct AI Personalities, each with specialized threat thresholds and communication styles.
- ✓ PhantomSentinel: Maximum security focus. Formal, zero-tolerance blocking.
- ✓ SPECTRE: Investigative. Focuses on attributing *how* an attack works to teach you.
- ✓ AURA: Encouraging. Prioritizes everyday analogies for non-technical users.
- ✓ Nyx: The Stealth Operative. Operates silently, alerting only for critical system events.