Reduce 9-1-1 emergency response times in a major city

Open by restating the problem and asking 3-5 clarifying questions: what is the baseline response time, what is the city trying to optimize (median or 90th percentile), what are the political constraints, who are the key users (dispatchers, EMTs, citizens), and what data is missing. Identify four pillars: (a) demand forecasting by hour and zone, (b) dynamic ambulance staging, (c) intelligent dispatch (which unit to which call, accounting for hospital capacity and traffic), and (d) feedback loop (post-incident review and continuous retraining). Sketch the entities (Call, Unit, Zone, Hospital), key events (CallCreated, UnitDispatched, Arrived), and at least two dashboards (a real-time ops view for dispatchers and an executive throughput view). For v1, scope to a single borough or a single hour of day and define success metrics (P50/P90 response time, fraction of calls with > 1 unit available). Spend the last 10 minutes naming failure modes (data latency, hospital diversion, surge events) and how the system degrades when an upstream feed goes stale.