In Maslow’s Hierarchy of Needs, safety is classified as essential for driving growth behaviours such as seeking employment and financial security. India’s vision of becoming a developed country by 2047 cannot be realised without a focus on safety and policing.
Staqu Technologies co-founder and CEO Atul Rai grew up in eastern Uttar Pradesh in an era when the region was plagued by crime, violence, and weak law enforcement. “When you grow up seeing murder and theft as common occurrences, you either prepare for the civil services, or you try to solve it another way,” he tells AIM.
Rai chose the latter. After studying AI in the UK and publishing research at international conferences, he returned to India convinced that AI could help address the country’s chronic shortage of policing resources. According to the India Justice Report 2025, India has roughly 155 police personnel per 100,000 people. “A society cannot be developed if it is not safe,” he says. “If we can help make safety accessible, not privileged, then we’re building something that matters.”
He decided to do something about it and teamed up with Anurag Saini and Pankaj Kumar Sharma to found Staqu Technologies in 2015. The startup was born from the belief that CCTV cameras, already proliferating across Indian cities, could be transformed from passive recording devices into active tools for intelligence, prevention, and accountability.
Making “Dumb Cameras” Intelligent
At the heart of Staqu’s offering is JARVIS, an AI-powered video analytics platform that interprets live and recorded camera feeds in real-time. During Staqu’s early years, artificial intelligence was still nascent, and the idea of letting AI, not humans, monitor video feeds was unconventional. But Rai sees it as essential.
“When humans watch cameras, intention creeps in. Bias creeps in. Misuse creeps in,” he highlights. “AI should monitor what’s happening. Humans should act only when needed.”
The Gurugram-based company’s earliest deployments were with police forces and government agencies, an unusual go-to-market strategy for a startup. Staqu engineers often worked on-site, sometimes for weeks, to understand how investigations actually unfolded on the ground. Over time, its systems were used to help identify suspects, track their movements, and correlate video evidence using attributes like clothing, vehicles, and behaviour patterns.
Staqu has worked with more than 11 state police forces and multiple intelligence agencies. Its systems reduce the time required to analyse large volumes of video data by enabling text-based searches across footage. Instead of manually scanning feeds, officers can search for “a red car” or “a person in a white shirt and blue jeans,” and retrieve relevant clips instantly.
This public-sector work shaped Staqu’s philosophy of building for complexity first. “If you can solve crime and security, then solving retail analytics is a cakewalk,” Rai stresses.
AI With Accountability
Operating at the intersection of surveillance, safety, and analytics calls for accountability.
The company deploys its software on customer-owned infrastructure rather than its own servers. Police deployments run within government data centres; enterprise deployments run on private clouds. Staqu is GDPR-compliant, uses end-to-end encryption, and undergoes regular security audits. Rai asserts the company prioritises data protection. Its operations align with SOC 2 standards, and it holds certifications from local cybersecurity firms. Data is encrypted and transmitted via an RTSP tunnel to a customer-owned server running JARVIS.
“Data privacy is very important in the world of AI. Data has to be secure, and there has to be accountability. What we did was deploy JARVIS, not on our own servers. We do not have our own cloud. When we deploy with the police, JARVIS runs in their data centres. When we deploy in the private sector, say with Raymond, we deploy on the client’s cloud,” Pai highlights.
In terms of data access, he mentions multiple layers, such as passwords, OTPs, and email-based authentication. “Unless someone from within the organisation intentionally shares access, the data cannot be misused. JARVIS itself cannot do that,” he asserts.
Building Intelligence Systems
Staqu’s most ambitious public-sector work is YAKSH, an advanced criminal intelligence platform built on top of JARVIS and deployed with several state agencies. Designed to unify fragmented criminal data, images, videos, audio samples, and text records, YAKSH enables multimodal reasoning across datasets, helping agencies identify patterns difficult to detect manually.
New inputs are matched against existing profiles to reduce duplication, while all insights are cross-verified to minimise AI hallucinations. The platform supports gang mapping, hotspot detection, and discovery of hidden associations, while retaining human-in-the-loop validation.
From Public Safety to Private Intelligence
Staqu’s move from B2G to B2B was prompted by a structural disadvantage that brick-and-mortar stores faced compared to e-commerce platforms: a lack of behavioural data.
One of Staqu’s most prominent enterprise deployments has been with Raymond Limited, which operates over 1,400 stores across India. Using JARVIS, Raymond gained real-time visibility into footfall, conversion, and engagement across regions.
The insights led to tangible outcomes. In several underperforming stores, analytics revealed that low conversion was not due to footfall, but weak customer engagement. Focused staff training led to conversion improvements of up to 30% in those locations, while campaign testing helped increase footfall for Raymond’s ethnic wear line, the startup says.
In hospitality, Olive Hotels uses JARVIS to operate properties remotely, combining video and audio analytics to monitor guest sentiment, SOP compliance, and service quality, resulting in a 50% reduction in manpower costs. The startup has also deployed its technology in large malls and mixed-use commercial spaces to analyse queues, congestion patterns, and dwell time, helping operators optimise staffing, security placement, and tenant performance. In manufacturing and enterprise campuses, the platform flags safety violations, restricted-area breaches, and process non-compliance in real time.
Staqu’s analytics are also deployed across food and quick-service restaurant chains in India and the Middle East, including Dunkin’. JARVIS helps brands understand peak demand windows, staff-to-customer ratios, and queue abandonment—data that traditional POS systems cannot capture.
Healthcare is another growing vertical. Staqu works with hospital groups in India and the Middle East to monitor patient flow, safety compliance, and operational efficiency across facilities.
Business Built on Restraint
Staqu has adopted a rather restrained funding strategy. Since its inception, the company has raised approximately ₹14.5 crore, including a pre-Series A round backed by SIS Limited and Mount Judi Ventures.
Staqu has relied heavily on revenue reinvestment to fund growth. The result is a rare profile in Indian AI: Staqu is EBITDA-positive, operationally profitable, and generates roughly ₹36 crore in annual recurring revenue. It operates on a pure SaaS model, charging customers on a per-camera, per-month basis, with pricing ranging from about ₹1,500 to ₹8,000, depending on the level of analytics. Most enterprise and government contracts run three to five years, providing predictable cash flows.
Rai is candid about the trade-offs. Building without abundant capital slowed early growth, but preserved control. “AI is not a business you can run on fixed plans,” he said. “The field changes too fast. Founders need freedom to adapt.”
That discipline has also shaped investor perception. In an interview with Indian Startup Times, Sherif Kottapurath of Mount Judi Ventures categorised Staqu as an “impact-neutral” investment, one that may not directly target social outcomes, but creates measurable value through applications in safety and efficiency.
An Indian AI Company By Design
Unlike many AI startups that look westward, Staqu has remained unapologetically India-first. Its models are trained on local contexts, its products shaped by domestic constraints, and its ambitions tied to national needs.
“We’re not building foundational models,” Rai highlights. “We’re building application-layer AI that actually gets used.”
Staqu aims for an IPO toward the end of the decade at a market valuation of ₹8,000 crore.
In India, Staqu Technologies competes with a growing number of companies in video analytics and surveillance intelligence, including Videonetics (smart cities and traffic), AllGoVision (modular behaviour and crowd analytics), Innefu Labs (government intelligence and law enforcement), and CRON Systems (perimeter security).
However, Staqu says it stands out by providing a single software platform that addresses both public-sector security and private-sector operational analytics, rather than focusing solely on smart-city or retail analytics.
The post Bird’s Eye: How Staqu Tracks And Solves Crime in India with AI appeared first on Analytics India Magazine.
