AI-Assisted Public Feedback Platform
Deploying an AI-powered platform to aggregate citizen feedback and route issues for rapid resolution.
Punjab launched an AI-assisted public feedback platform to aggregate input from helplines, WhatsApp, and service counters. The system classifies, deduplicates, and routes signals to responsible teams for timely action.
The goal was to shorten the time between a citizen raising an issue and the government acting on it—while creating visibility into systemic problems that require upstream fixes.
What we put in place
- Unified intake across helpline, WhatsApp, web forms, and in-person counters
- Entity resolution and deduplication to avoid double-counting
- AI triage to tag department, location, severity, and topic
- SLA-aware routing to district and department focal persons
What made it work
- Tight handoffs between contact centers and field teams
- Human-in-the-loop review to improve model labels
- Dashboards that surfaced aging tickets and bottlenecks
- Weekly reviews to remove structural blockers
We prioritized a minimal viable workflow: capture, classify, assign, resolve. As confidence grew, we layered analytics for topic drift, repeat locations, and seasonal spikes to inform upstream fixes.
Within the first quarter, median resolution time decreased by 37%, and first-contact resolution improved as guidance libraries grew. More importantly, recurring issues informed policy tweaks and preventive actions.
