Projects
AI-Enabled Grievance Redressal
Grievance redressal systems generate continuous, population-scale signals about the quality of public service delivery.
In Odisha, millions of complaints produce vast volumes of structured and unstructured administrative data each year—data that has historically been used primarily for case resolution, leaving its broader intelligence value untapped.
DPIC, in partnership with the General Administration & Public Grievance (GA&PG) and Electronics & IT Departments, is working to change this. By building structured data pipelines, testing applied AI tools, and developing analytical systems, DPIC is helping transform Odisha’s grievance infrastructure into a strategic platform for evidence-informed governance.
Building a Structured Grievance Data Foundation
This initiative draws on one of the state’s most significant administrative data assets: 1.4 million structured grievances with complete action histories, nearly one terabyte of linked unstructured documents and attachments, and detailed resolution timelines and service-level metadata.
DPIC has designed and implemented privacy-preserving data pipelines to clean, standardize, and link these datasets—creating a unified, machine-learning-ready grievance data foundation. Interoperability with related service delivery systems such as the Odisha Right to Public Services (ORTPS) and Ama Shasana feedback surveys enables cross-domain analysis while maintaining grievance redressal as the core analytical unit.
Testing and Validating AI Tools for Grievance Workflows
A significant share of grievance data arrives in unstructured forms—scanned letters, handwritten submissions, and document attachments. DPIC is developing and rigorously evaluating AI tools, including natural language processing (NLP), optical character recognition (OCR), document intelligence, and automated PII redaction, to assess their value within grievance processing workflows.
These tools, when validated, enable structured extraction from scanned submissions, privacy-compliant redaction of sensitive citizen information, and automated classification and routing of incoming grievances. DPIC’s approach emphasises careful piloting and structured performance evaluation—testing accuracy, efficiency gains, and citizen-level impact at small scale before any consideration of broader deployment.
From Complaint Data to Governance Intelligence
Beyond workflow improvements, DPIC is building an analytical engine to convert grievance data into decision-support intelligence. By integrating structured and unstructured datasets, the system is designed to enable:
- Early detection of systemic service delivery failures before they escalate.
- Identification of grievance surges linked to specific policy changes or administrative events.
- Predictive signals of emerging stress points across districts and departments.
- Equity analysis of complaint patterns and resolution rates across demographic groups.
The objective is to shift Odisha’s grievance system from reactive complaint resolution to anticipatory, intelligence-driven governance where patterns in citizen feedback actively inform how services are managed and improved.
Better Governance Through Grievance Intelligence
Grievance systems generate continuous signals about service delivery challenges, and when properly structured and analytically interrogated, they become something more: an early-warning system for administrative failures, a diagnostic for inequitable service delivery, and a foundation for continuous improvement.
By integrating large administrative datasets with applied AI tools, Odisha is strengthening its ability to unpack complex information embedded in grievance records, anticipate recurring citizen concerns, and respond proactively rather than reactively. This approach builds the state’s capacity to understand citizen experience at scale—reducing administrative burden while enabling more timely, evidence-informed governance decisions.