Wizard of Oz

AI Service Economics & Latency Simulator

SF Startup $6,000 Ad Spend Experiment vs LLM API Queue Dynamics
Human Cost / Query
$1.85
65 WPM | 12h Shift
LLM Cost / Query
$0.012
$0.002 / 1k Tokens
Max Queue Backlog
342 chats
Overflow at Hour 4
Founder Fatigue Index
98%
Severe Burnout Hazard
Queue Overflow Hour
Hour 4
Traffic > Human Throughput
Live Discrete-Event Message Queue Visualization Simulating User Inflow vs Typing Pipeline
Incoming User Query
Human Queue Backlog
LLM Inference Queue
Resolved Chat
Hour 00:00
Queue Backlog & Surge Traffic (24h)
Latency Distribution (Human vs LLM)
Canonical Experiment Proof & Operating Point
Human Cost / Query: $1.85
LLM Cost / Query: $0.012
Max Queue Backlog: 342 chats
Founder Fatigue: 98%
Queue Overflow Hour: 4
The $6,000 ad experiment generates ~4,000 active chats over 24h. Human typing speed (65 WPM) caps operator capacity at ~24 chats/hour per operator shift. Queue backlog breaks at Hour 4, causing founder fatigue to spike to 98% and per-query human costs to surge to $1.85 vs $0.012 for LLM API.
Enjoy this tool? Build your own with Super