Sentinels is our global event intelligence platform — the project that started Global Ring AI. At its core is a real-time anomaly detection system that processes enriched event data continuously, looking for five distinct signal types that indicate something worth human attention.
The Five Detectors
Each detector monitors actor-level activity streams, maintaining state and historical context to distinguish meaningful change from background noise:
Volume Spike detects sudden increases in mention frequency for a given actor. It maintains a rolling baseline and triggers when current volume significantly exceeds the norm. A head of state who normally appears in a handful of articles suddenly appearing in hundreds is a different kind of signal than one who’s always in the news.
Silence Break is the inverse — it watches for actors who have been quiet and suddenly reappear. Extended silence followed by activity is often more significant than sustained high volume. When someone who hasn’t been mentioned in months suddenly surfaces, that’s worth investigating.
Tone Shift monitors sentiment trajectories. A gradual drift in tone is noise; a sharp inflection is signal. The detector uses windowed sentiment analysis to distinguish the two. An actor whose coverage shifts from neutral to sharply negative in a matter of hours is exhibiting a pattern that warrants attention.
New Connection fires when an actor appears in a relationship context for the first time — a new co-mention, a new organizational affiliation, or a new geographic association. Novel connections between previously unrelated actors can signal emerging alliances, conflicts, or coordinated activity.
Coordination Signal looks for patterns that suggest coordinated activity: multiple actors exhibiting similar behavior changes within a narrow time window. When several actors in a network all shift tone or volume simultaneously, it may indicate an orchestrated campaign or a shared external catalyst.
Why Five Signals
No single detector captures the full picture. Volume alone misses the quiet actors. Tone alone misses the ones generating noise without sentiment. New connections alone miss the established actors whose behavior changes. The five detectors work together to create a composite awareness that approximates what a team of human analysts would notice — but across thousands of actors simultaneously.
The Human Layer
The detectors surface anomalies. They don’t make judgments. Every anomaly is routed to a human analyst workflow where it gets context, gets prioritized, and gets a decision. The AI handles the scale problem — watching thousands of actors simultaneously. The human handles the judgment problem — deciding what matters and what to do about it.
This is the augmentation pattern at its most concrete: machines watching everything, humans deciding what’s important.