Fast Instincts (<15ms • $0) for Slow Reasoning LLMs
System 1: ActiveSystem 2: Idle
Decision Inputs
0.80
If confidence falls below this threshold, Reflex auto-escalates to System 2.
Dual-Brain Architecture TriageReal-time Telemetry
🧠 System 1: Spinal Reflex
Latency: 0.08 msCost: $0.000000
RESOLVED
Noul Epistemic Probability0.9620
Choice Distributionquarantine
⚡ Evaluated in single machine-native logit pass without LLM autoregression.
🏛️ System 2: Cerebral Cortex
Latency: 0 msCost: $0.0000
IDLE
Deliberative Reasoning Trace
System 2 is IDLE.
Confidence (0.96) exceeded epistemic gate threshold (0.80).
Reflex resolved this request at the spinal cord, saving 1,800ms of latency and $0.03 in Claude/GPT API costs.