Aurm’s 24/7 accessible locker system is not just a safe, it’s a sensor-infused, AI-powered strong room designed to think and act on its own. In a candid conversation with Express Computer, Vijay Arisetty, Founder & CEO, and Pratap Chandana, Co-Founder & CTO, break down how Aurm combines reinforced physical infrastructure with real-time telemetry, anomaly detection from surveillance feeds, and automated incident response to ensure round-the-clock security—even without human supervision. From filtering out false alarms like tyre bursts to triggering virtual war rooms in under 15 minutes, the duo explains why they believe legacy security systems are reactive and how Aurm is pushing toward a more predictive, data-driven future.
What we observed is that the way Indians live is undergoing a fundamental shift. Earlier, people lived in individual homes spread across layouts. Now, there’s a move toward vertical living in gated communities, urban clusters that are densely packed within a few pin codes. These new developments, like Bellandur, Electronic City, Whitefield, or Sarjapur, are seeing a growing concentration of affluent residents.
Vijay Arisetty: However, the traditional banking model hasn’t evolved accordingly. Banks operate under formulas that mandate a fixed number of branches or lockers per square kilometre, which doesn’t account for the density in these areas. As a result, there’s a clear mismatch between the demand and availability of bank lockers. People still need secure spaces to store physical valuables, but lockers are increasingly scarce.
When we studied the problem deeper, we saw an opportunity. Instead of trying to find the right customer for a product, here, the customers were already coming together in one place, gated communities. We realised that if the market is self-organising, we should build the solution right there, integrated within these communities.
But we also knew we couldn’t replicate the traditional bank locker model—it’s too manual, too dependent on physical keys, and not scalable. That’s when we decided to reimagine the solution entirely. We envisioned a smart, AI-powered, fully automated locker system that could be monitored and controlled remotely, offering stronger security than anything currently available.
It took us nearly two and a half to three years to build this. In the beginning, only a few of us believed in the idea—my co-founder Suraj and Pratap were among the first to join me. I had to paint a clear vision to bring them on board. Once we started building, what we initially imagined to be a certain scale turned out to be 10X better. We invested heavily not just in the tech but also in the overall user experience, safety, and security protocols. We’re dealing with people’s physical wealth, so like banks, we had to build trust through rock-solid infrastructure. We’ve made it even harder than traditional systems for anyone to compromise the security.
That’s how the idea for Aurm evolved, from a consumer insight about density and locker scarcity to a tech-first, scalable, secure platform integrated right into the heart of where people live.
Currently in how many locations do you have this facility set up?
Arisetty: At this point we are in 13 locations in Bangalore. As we speak, 7 more are being set up in Hyderabad and Visakhapatnam.
A few days ago, there was news about a partnership with Prestige involving AI-powered lockers. Could you share more details about that?
Arisetty: When we developed this product, a lot of builders realised that it addresses a fundamental need for residents. They started asking, “Why not offer this as an amenity?” That’s how some of the leading builders, starting with Prestige, partnered with us to provide this feature as a built-in amenity in the apartment premises for residents. This is how the partnership came about.
Can you elaborate on how AI is embedded in your system—not just for access control, but in predictive maintenance, anomaly detection, or even customer experience personalisation?
Pratap Chandana: AI is definitely going through a maturity curve, and while everyone’s talking about it now, we’ve actually been using its precursors—like machine learning—for quite some time. Essentially, we train our systems on large volumes of sensor data, which allows us to predict outcomes and drive efficiency.
For instance, predictive health of equipment is a key use case. Our promise to customers is 24×7 access. People might need to access their belongings at odd hours—say, 4 AM to attend a wedding. Any failure at that moment would be unacceptable. So, our systems constantly collect real-time data from nearly 2,000 sensors installed across our storage units. This data helps us identify and predict issues—be it mechanical, electronic, or sensor-related—before they happen, allowing us to proactively replace or fix components and ensure uninterrupted access.
We also use AI in access control. Initially, we implemented time-based OTPs (like Google Authenticator) for secure logins. But to enhance convenience, we’ve moved toward biometric solutions like palm vein recognition. It’s like helping someone take a passport photo by analysing selfies—they don’t need to pose rigidly; the system intelligently adjusts and verifies.
Another area is anomaly detection. Our sensors constantly collect environmental data, and over time, we’ve trained AI models to distinguish real threats from background noise. If earlier we responded to 10 alerts only to find they were all false positives, the system now filters such noise, letting us respond only to meaningful alerts.
We’ve also extended AI to our surveillance systems. Instead of relying on humans to monitor live feeds, we use AI to detect unusual activity inside strong rooms. For example, we track temperature and humidity data to identify conditions favorable for termites. When such conditions persist, we preemptively carry out pest control, often a week in advance.
Unlike traditional bank lockers, Aurm lockers are accessible 24/7. From an operational standpoint, how do you ensure secure, real-time authentication and activity logging at odd hours without human supervision—especially in densely populated residential complexes?
Chandana: From an operational standpoint, our system is designed to be largely self-healing and programmatic. If the system detects an issue like a malfunction or irregularity it automatically resets itself and resumes normal operations. In rare cases where manual intervention is needed, we can trigger a remote restart, but our goal is to solve most issues autonomously in real-time.
Regarding security and monitoring, we rely heavily on telemetry and behavioural pattern analysis. While we don’t collect user-specific data beyond what’s needed to deliver the service, we do monitor machine-level signals and system behaviours to detect anomalies. Traditional systems tend to be reactive; we aim to be predictive.
We perform anomaly detection using both sensor and surveillance feeds. For example, at the end of the day, our system analyses 24-hour video footage using AI algorithms to identify any unusual activity that might have been missed during live monitoring. This helps ensure nothing slips through the cracks. AI is being gradually infused throughout our system not as a standalone feature, but as a foundational layer that enhances reliability, especially during odd hours and without human supervision.
What about physical intrusion? The system is tech-heavy, but you also need physical strength in your infrastructure, right?
Arisetty: Absolutely. It’s like maintaining your health, you may go to the gym and build strong muscles, but you also wear smartwatches, glucose monitors, and other devices to collect data and take preventive actions. Similarly, our system blends physical strength with intelligent monitoring.
We’ve built strong rooms using steel and concrete, infused with sensors. These detect, prevent, and anticipate intrusion attempts. For instance, if someone is digging even 100 meters away, the vibration is picked up by our sensors. We analyse such data to determine whether it’s a real threat.
Once, an alarm was triggered due to a truck’s tire burst near a community. It sounded like an explosion, so our system flagged it. After investigation, we identified it as a false alarm. Now, the system knows how to recognise tire bursts and lowers the severity for similar incidents in the future. All this makes our system smarter and safer over time.
What happens automatically if an alarm goes off?
Arisetty: The moment an alarm is triggered, several things happen via APIs:
The security guards are immediately informed and dispatched to the clubhouse where the lockers are. Local police are notified about the intrusion attempt. We’ve hired Quick Reaction Teams (the same ones used by ATMs and banks). They’re alerted instantly and reach the site within 15 minutes. Our management and operations teams also rush to the spot. All of this is coordinated and executed in under 15 minutes.
Do you have plans to expand into Tier 2 or Tier 3 cities?
Arisetty: Not at this stage. We already have sufficient demand in Tier 1 cities. We’re focusing on areas with dense urban populations. However, a Tier 2 city like Visakhapatnam has shown strong demand in a few communities, and therefore we are only considering it.