Core Internet Immune System architecture
Complete biological immune-system implementation across innate, adaptive, and memory layers — combined in a single platform.
All 14 filings list Jeffrey Grayson as first named inventor. Nine cover the AegisQ Security architecture (filed August 2025, expiring August 2026); five cover AI Sentinel (filed April 2026). Two Security applications are slated for non-provisional conversion (Priority 1); the remainder pursue trade-secret protection or extended provisionals.
Complete biological immune-system implementation across innate, adaptive, and memory layers — combined in a single platform.
Peer-to-peer threat-intelligence sharing across tenant boundaries. The "immunity propagation" mechanism that makes new detections propagate within minutes.
The cross-tenant write-back mechanism that turns one customer's incident into every customer's defense.
XGBoost + Random Forest 70/30 ensemble with biological labeling; the detection model that achieves 99%+ accuracy.
Vertex AI–based continuous training loop with federated update mechanism across tenant datasets.
Multi-region Pub/Sub backbone with BFT consensus for cross-region threat-event distribution.
Post-quantum blockchain-based immutable detection-history storage; H2 2026 production roadmap.
Scoped delegation system (view_alerts, manage_alerts, respond_to_threats, full_access) for managed-service provider access.
Unified write-back interface to 40 security vendors (Cloudflare, CrowdStrike, Okta, Palo Alto, …) for sub-100 ms autonomous containment.
Declarative spec for what an AI agent is allowed to do, spend, and access. Charters are the source of truth for every detection and response decision.
Unified proportional-response taxonomy spanning observation, alerting, throttling, intervention, and emergency halt — applied identically to LLM agents and physical robots.
Pre-execution validation of joint limits, velocity, force, proximity zones, and geofences — enforced in real time across collaborative arms, AMRs, humanoids, and drones.
Time-limited cryptographic capabilities that let robots continue safe operation during connectivity loss; expiry triggers hardware failsafe.
ML-DSA-65 (FIPS 204) signed compliance reports proving an AI system operated within charter — tamper-evident against quantum and conventional threats.
Patent details, technical specifications, and freedom-to-operate analysis available under NDA.