docs/PITCH.md

SentinelAI — One-line positioning

Do not say: “AI cybersecurity dashboard.”

Do say:

SentinelAI is an autonomous multi-agent AI Security Operations Center that continuously monitors infrastructure, correlates attacks, performs AI-driven investigations, and generates remediation workflows using AMD-accelerated open-source AI infrastructure.

30-second version

• Autonomous collectors tail auth and demo logs (extend to Docker/K8s/cloud).
• Detection — brute-force, privilege abuse, malicious-source hits.
• Correlation — timelines and graphs, not isolated alerts.
• AI analyst — enterprise narratives, severity reasoning, action list (Ollama / vLLM / ROCm).
• Command deck — live feed, replay, ROCm panel, agent activity.

Why judges care

Angle	Hook
Autonomy	Runs without chat prompts; event pipeline drives agents.
Multi-agent	Collectors, detection, correlation, analyst, remediation, alerts.
Data residency	Local open models — no mandatory third-party LLM API.
AMD	ROCm = throughput for concurrent SOC reasoning at hackathon cost.

Closing

“We’re not a chatbot bolted onto logs — we’re a mini-SOC runtime you can demo in minutes and extend to real auth.log and TI feeds.”