110 Smart Police Dispatch Integrated Platform

Project Background

A city 110 command center handles 4,000+ daily emergency calls. Under the traditional call-taking model, operators had to listen and take notes simultaneously while filling out forms manually, with an average call-taking time of 3-5 minutes, severe queuing during peak hours, a missing-field rate as high as 15%, and address accuracy of only 70%. Chinasoft International responded quickly through the AI BootCamp training format, assembling an FDE combat team (5 members) to complete the full process from scenario diagnosis to system delivery in 3 days.

Solution

Extreme delivery: completed a complete system delivery of 158 files and 27,263 lines of code in 3 days, with a front-end/back-end separated architecture integrating six external services including Volcano ASR/TTS, JointPilot Lingxi, and Amap.
Core metrics improved: call-taking time reduced from 3-5 minutes to 28 seconds (↑84%), missing-field rate dropped from 15% to 0.8% (↑93%), and address accuracy increased from 70% to 96% (↑37%).
AI-native capabilities: full-process AI-driven with token consumption throughout, enabling real-time voice transcription, intelligent element extraction, incident-tag classification, and intelligent dispatching and transfer, with 100% structured data.

Customer Testimonial

This training completely refreshed my understanding of AI agents. Through three days of hands-on practice, I was deeply involved in the entire project process, leveraging AI to efficiently tackle challenges.