Gurugram: Haryana Chief Minister Nayab Singh Saini on Monday inaugurated the Haryana Artificial Intelligence (AI) Sandbox at a high-profile event in Gurugram.
The launch, held in partnership with the World Bank, was part of the state’s larger Haryana AI Development Project (HAIDP), a three-year programme running from 2025 to 2028 with a total outlay of Rs 474.39 crore, of which the World Bank is putting in 70 per cent and the Haryana government the remaining.
Here are a few examples of what the Sandbox solutions would look like in practice for an ordinary citizen.
Take the Jansamvad grievance platform of the Haryana government. Every month, tens of thousands of complaints pour in—pensions not received, electricity connections pending, water supply disrupted, roads broken, police matters unresolved.
A significant amount of official time goes not into solving these complaints but simply into sorting, routing and deduplicating them. Many are wrongly marked. Many are the same complaint filed multiple times out of desperation.
An elderly person from Hisar writing that his old age pension has not arrived in three months and that he has complained several times before—that complaint needs to be read, understood, marked urgent, linked to similar complaints from the same district, and sent to the right department. The AI tool being tested does exactly that, automatically, across thousands of complaints simultaneously.
The officials then focus on solving problems, not on sorting papers.
आज हरियाणा AI सैंडबॉक्स का लोकार्पण किया।
यह मंच विभाग की चुनौतियों के लिए नए तकनीकी समाधानों को विकसित करने और उनकी प्रभावशीलता परखने का अवसर प्रदान करेगा। pic.twitter.com/qCnYR0xBHh
— Nayab Saini (@NayabSainiBJP) June 1, 2026
Take another example. A citizen photographs a pothole on the Gurugram-Sohna road and uploads it through a mobile application. The AI reads the image, assesses the severity of the damage, tags the location, and checks whether fifty other complaints have arrived from the same stretch that week. If they have, the system flags that corridor as a priority repair before an accident forces the issue. Two hundred complaints from one road in a week would trigger an alert to engineers. The idea is to move from reactive repairs to anticipatory maintenance.
The engineer still approves the work order. The AI just ensures the worst roads do not wait behind less urgent ones in the queue.
The AI does not replace the official. It reads what the official does not have time to read, spots what the official cannot see across thousands of data points, and puts the most urgent cases at the top of the pile.
Powering AI Sandbox
The man who has been building it from the inside is J. Ganesan, a 2006-batch Haryana IAS officer who serves as Managing Director of HARTRON, the Haryana State Electronic Development Corporation Limited, the nodal agency for the project, besides holding several other important posts.
Contacted by ThePrint on Monday, Ganesan said that last year, the Haryana government launched ARJUN (AI for Resilient Jobs, Urban Air Quality and Next-Gen Skills), a Special Purpose Vehicle (SPV) formed to execute World Bank-aided projects.
“The Haryana CM is chairman of the SPV, and I am its chief executive officer. The SPV is already implementing the Haryana Clean Air Programme for Sustainable Development, a Rs 3,500-crore Haryana Clean Air Project undertaken in collaboration with the World Bank. The programme focuses on industrial emission control, promotion of electric vehicle adoption, and agricultural interventions to improve environmental standards.
By integrating AI with environmental and sustainability goals, Haryana is leveraging technology not only for economic growth but also for climate resilience and improved urban air quality.
A second project, the Haryana AI Development Project, is a Rs 474-crore state-led initiative, backed by a 70:30 funding partnership with the World Bank. Its primary goal is to establish Haryana as a premier national hub for artificial intelligence by focusing on education, startups, and public governance. It was during the implementation of these projects that the World Bank offered us the Haryana AI Sandbox,” said Ganesan.
He added that a team from the World Bank is constantly involved with Haryana officials for the development and implementation of the Sandbox.
Fail cheap, learn fast
In technology, the word ‘sandbox’ has a specific meaning. It refers to a controlled, isolated environment where new software or systems can be tested without affecting anything in the real world.
Think of it as a rehearsal space for a theatre production, where mistakes do not bring down the curtain on opening night. An AI Sandbox applies this principle to government.
Governments collect enormous amounts of data, on crop yields, hospital admissions, student performance, air quality, water levels, but rarely have structured mechanisms to safely experiment with that data to build solutions.
The risk of getting it wrong when dealing with public services is too high to simply try things out in the open. Farmers, patients and schoolchildren cannot be guinea pigs.
A sandbox provides a controlled innovation environment with secure data-sharing, governance safeguards, and technical support, allowing governments and innovators to validate solutions before full-scale adoption.
A startup or a research team brings an AI-based tool, tests it against real government data in a safe, ring-fenced environment, with computing resources and datasets provided by the government. If the solution works, it is scaled up. If it does not, the failure stays contained and costs nothing at the delivery end.
Ghost in the machine
When Saini arrived at Taj, Gurugram for the afternoon launch, Ganesan was asked to give a demonstration. He chose not to open with a slide deck or a technical walkthrough. He opened with a name. Jai Bhagwan. A resident of Bohar village in Rohtak.
“When Jai Bhagwan reached the age at which he became eligible for an old age pension, he applied. His application entered the system at a moment in life when a person needs the maximum sensitivity from the state — when income has dried up, when the body is slowing down, when the margin for waiting is the thinnest. His file moved from one office to the next. Back and forth, up and down the bureaucratic chain. Before any action could be taken, Jai Bhagwan died,” Ganesan told the gathering.
Ganesan said the incident does not just indict a process. It shatters the conscience of everyone involved, including the officers. Because behind every file, he said, there are not just papers — there are human beings, their families, their lives.
What makes the story more pointed is where it surfaced. It did not come to light during an audit, the conventional mechanism through which government systems catch their own failures. It emerged during a review of the CM Window—the state’s digital grievance platform.
“Today’s launch of the AI Sandbox is an attempt to ensure that no more Jai Bhagwans are lost in the system in future,” said Ganesan.
It was, in its way, a more effective argument for Artificial Intelligence in government than any technical presentation could have been.
Five problems, five solutions
Ganesan told The Print that Haryana has identified five specific “use cases”—real, existing government problems—for which solutions will be sought through the Sandbox.
The five are: complaints triaging and routing for Jansamvad; roads defect detection and prioritisation for the Mhari Sadak scheme; teachers deployment planning and resource optimisation; radiology review prioritisation and screening support; and water grievances clustering and infrastructure-linked prioritisation.
These are not abstract AI experiments. Each maps directly onto a chronic pain point in Haryana’s public administration. Jansamvad is the state’s public grievance platform, which receives thousands of complaints — the challenge is routing each complaint to the right department fast and accurately, something that currently depends on human intermediaries and suffers predictable delays.
The Mhari Sadak scheme involves road maintenance across the state; an AI-based defect detection tool using camera footage or satellite imagery could flag potholes and damage before they become accidents.
Teacher deployment is a perennial problem in Haryana — some schools are overstaffed while others run on a single teacher; an AI tool analysing enrolment patterns, dropout rates and geography could enable rational posting decisions.
Radiology is a bottleneck in district hospitals where radiologists are few and the queue is long; AI screening tools can flag priority cases. And water grievances, often localised and repetitive, can be clustered geographically to identify infrastructure failures rather than treating each complaint in isolation.
“Basically, today what we are doing at Gurugram is that we are giving these five identified problems to the startups. We will provide them all digital help, computing, data, etc. from the government and ask them to come up with the solutions. The government will then implement the best solutions that come up,” said Ganesan.
(Edited by Nardeep Singh Dahiya)
Also Read: Years after CAG raised red flag, Haryana blacklists ghost publishers who scammed govt colleges