Operationalization of an AI Sandbox for responsible AI adoption across public-sector departments in Haryana

Subscribe
FR
Light Dark

Operationalization of an AI Sandbox for responsible AI adoption across public-sector departments in Haryana

Aug 17, 2026
2 min

Artificial intelligence (AI) can help governments improve public services. Yet, its use in public systems requires more than promising technology. Governments need a safe way to test if solutions are accurate, useful, secure, and accountable before they affect citizens or government decisions.

The World Bank-supported AI Sandbox has created one such pathway in the state of Haryana in India. The sandbox provides a secure, governed environment where AI solutions can be built, tested, and validated for real government challenges using approved, controlled, or representative public-sector data.

Located in northern India near Delhi, Haryana has active digital public systems and service-delivery workflows in which AI can support improved classification, prioritization, risk detection, and decision-making. MSC (MicroSave Consulting) worked with the World Bank and the Government of Haryana to identify opportunities for AI adoption. The MSC team engaged with government departments to understand their operational challenges, review available datasets, assess data readiness, and translate these needs into structured challenge use-case briefs for innovators.

The identified use cases span multiple public-service domains. In education, AI can support teacher deployment planning and resource optimization. In healthcare, it can assist with TB screening and radiology-based decision support. In citizen services, AI can improve grievance classification, routing, and response management through Jansamvaad, Haryana’s citizen grievance platform. In infrastructure and utilities, it can support road defect detection through Mhari Sadak, Haryana’s road complaint and monitoring platform, and strengthen water quality risk intelligence for the Public Health Engineering Department.

Following the challenge design phase, where use cases were refined into clear problem statements, data requirements, and application expectations, the Sandbox will open for applications from startups, technology providers, research institutions, and consortiums. MSC will support the development of application guidelines, evaluation criteria, and screening processes to identify solutions that show strong problem fit, technical feasibility, data readiness, and responsible AI practices. Teams selected through this process will then build and validate their solutions in the Sandbox using approved data and defined testing criteria, with successful solutions progressing toward pilot implementation.