Around the world, AI has already shown promise in predicting crime patterns. For instance, Raven Sentry, an American AI tool, achieved 70% accuracy in anticipating Taliban attacks in Afghanistan by 2020. Learning from such successes, Maharashtra became the first Indian state to create an independent AI-driven law enforcement unit.

MARVEL is a Special Purpose Vehicle (SPV) formed through a tripartite agreement between the Maharashtra government, IIM-Nagpur, and Pinaka Technologies. It focuses on building tailored AI tools for specific policing needs, from tracking missing persons to solving organised crime cases.

“Our mission is to augment—not replace—human policing,” says Harssh Poddar, SP Nagpur Rural and MARVEL’s CEO. “Law enforcement generates massive volumes of data. AI can analyse patterns humans might miss, enabling faster and more accurate decisions.”

How AI augments crime prevention

With urban areas blanketed in CCTV cameras and digital devices, many crimes now leave a digital trail. MARVEL’s AI algorithms are designed to analyse this vast visual and metadata, detecting patterns that aid both crime prevention and post-incident investigations.

Take chain snatching, for example. AI can pick up minute details in CCTV footage—like the make of a motorcycle, helmet type, or travel speed—that might escape human notice. For hit-and-run cases, AI can scan toll plaza cameras and calculate which vehicles passed the area at relevant times, narrowing suspects from hundreds to a handful.

In missing persons cases, especially involving women and children, AI rapidly processes smart city CCTV networks to trace last-seen movements. Even in complex organised crimes—such as narcotics, human trafficking, or wildlife trade—AI can map years of financial transactions, communication logs, and surveillance data to reveal hidden networks.

Building tools, not solving individual cases

Unlike an investigative agency, MARVEL doesn’t directly work on individual cases. Instead, it builds customisable AI solutions that police departments can deploy. For example, if a district police team needs a tool specifically for missing persons, MARVEL develops and hands over the required algorithm.

“Think of MARVEL as a technology enabler,” explains Poddar. “We design predictive models that highlight trends, and officers interpret the results for actionable insights.”

Currently, MARVEL is working on five AI projects for different agencies, supported by a team of 25 technical experts.

Scaling AI’s role in policing

While AI’s predictions improve with more data exposure, the technology is still evolving. Just as Raven Sentry grew more accurate in Afghanistan, MARVEL’s long-term goal is to refine its models over time, achieving near-flawless support for Maharashtra’s police forces. In April 2025, the Maharashtra government partnered with Microsoft to establish AI centres in Mumbai, Pune, and Nagpur to enhance governance using frontier technologies. 

“AI will not replace human judgment,” clarifies Sujata Saunik, Chief Secretary of Maharashtra. “But it will accelerate investigations, improve conviction rates, and free up officers for more critical fieldwork.”

By building this dedicated AI hub, Maharashtra aligns with India’s digital governance and smart policing priorities, setting a replicable model for other states. As MARVEL matures, it promises a future where human expertise and AI work hand in hand to keep citizens safer.