🤓 Latest Submissions
GateTrack Sentinel
GateTrack Sentinel is a human-governed visitor-risk operations platform designed for regulated and security-sensitive environments. It validates each visitor request, blocks unsafe or manipulative input before model inference, applies deterministic risk scoring and routing, and retrieves controlled read-only policy context. Fireworks AI is then used only to produce a bounded advisory narrative. The model cannot change the authoritative risk score, route, selected policies or final human decision. Every workflow stage is captured through bounded-loop contracts that record permitted tools, verification results, attempt limits, stop reasons and audit events. Each completed case can generate a portable SHA-256 proof packet containing the case record, decision lineage, source-confidence map, runtime evidence, loop-control records and a tamper-evident audit chain. The public demonstration uses synthetic data only. It demonstrates routine visitor clearance, escalated human review, prompt-injection blocking, structured live-provider reviews, deterministic replay and independently verifiable evidence. GateTrack Sentinel is designed for security, compliance and audit teams in regulated SMEs, casinos, DNFBPs and other organisations that need explainable and accountable AI-assisted operations. Its commercial pathway includes containerised subscription deployment, implementation support, policy configuration and compliance services through AUREX SICOS Advisory Ltd.
12 Jul 2026
