JARVIS transforms CCTV into intelligent surveillance with AI-powered audio and video analytics.
Founded in 2015 by engineers Atul Rai, Anurag Saini, and Pankaj Sharma, Gurgaon-based Staqu Technologies specialises in intelligent audio-video analytics, turning traditional CCTV cameras into AI-powered surveillance and insight-generation tools.
Its flagship product, JARVIS, acts as a real-time video and audio analytics engine capable of extracting actionable insights from standard CCTV feeds without requiring any new hardware. Connecting an existing camera’s IP address to the JARVIS portal enables instant monitoring and analysis.
While describing its features, Atul explains that JARVIS carries out both security and operational functions across multiple domains. In security scenarios, it can detect incidents such as violence, intrusions, fires, and overcrowding. Additionally, it performs activity recognition by analysing context-based behaviour over time, enabling it to identify events such as physical altercations or monitor routine tasks, such as cleaning.
For operational or commercial applications, such as in retail, hospitality, and co-working spaces, JARVIS generates insights including footfall count, gender ratio, customer journey mapping, area engagement analysis, queue times, and conversion rates.
While explaining the origin of the company’s name, Atul shares that, as engineers, they wanted something rooted in computer science. ‘Stack’ and ‘queue’ are two fundamental data structures, so they combined the two forming ‘Staqu’ as a nod to its technical background.
On the electronics front, Atul explains that JARVIS operates using a suite of AI models. It utilises convolutional neural networks (CNNs) for visual detection and incorporates advanced transformer-based models, including large vision models (LVMs), for classification tasks. On the audio side, JARVIS employs models for both speaker and scene recognition.
“The system is smart enough to match individuals across multiple cameras from different angles, even when the face is not visible, using gait analysis and body silhouette. In large events, such as religious gatherings in Ayodhya, JARVIS has been successfully deployed for crowd density monitoring, identifying fake vehicle plates, and conducting facial and reverse facial recognition for law enforcement,” he adds.
All video and audio data are analysed either on the client’s on-premise servers or private cloud infrastructure, adhering strictly to cybersecurity standards such as GDPR and ISO certifications.
The business model is based on a SaaS subscription. Clients pay on a per-camera basis, ranging from ₹2000 to ₹8000 per month, depending on the features and use cases.
Atul shares that Staqu currently supports over 170 enterprise clients across various sectors, including law enforcement, manufacturing, and retail. Notable clients include Ray-Ban, PVR, Porsche, BlackBerry, Embassy Group, and multiple state police departments across India. Deployments range from major metropolitan areas, such as Delhi and Mumbai, to smaller Tier-2 and Tier-3 cities, including Jaunpur and Mirzapur. Internationally, Staqu operates in nine countries and is expanding into new markets, including the US and the UAE. A Dubai office is underway, and future plans include establishing an R&D lab in academic hubs such as Cambridge or Oxford.
The startup reports that JARVIS handles over 400,000 image frames per second from thousands of cameras, with sub-second analysis latency. Staqu has published more than 25 research papers and holds two patents: one for real-time large-scale video frame analysis and another for privacy-preserving person re-identification technology that does not rely solely on facial recognition.
Benchmarking on standard datasets, such as LFW2 for visual tasks and VGG VoxCeleb for audio, shows accuracy levels between 98% and 99.7%, outperforming even some leading academic institutions.
The mobile accessibility of JARVIS enables users to monitor live streams, receive alerts, and query data through a dedicated app.
Atul notes that Staqu has achieved significant growth, with ARR rising from ₹25-30 million in 2022 to around ₹400 million in 2025, multiplying about 15 times. The company reports zero customer churn and maintains a 6% EBITDA profit margin, notable in the AI sector.
The company plans to expand into the B2C market within the next year, aiming to democratise access to AI-powered surveillance for individuals and smaller businesses.
