- Nikhil Teja Kolli's startup MokSa.ai uses AI-enabled security cameras to curb theft and fraud.
- Since its launch a year ago, MokSa.ai has garnered over 70 customers and $240,000 in ARR.
- MokSa.ai has secured $1.5 million in pre-seed funding and aims to expand its operations.
When he was a college student in Kansas, Nikhil Teja Kolli worked a night shift at a convenience store on the edge of the Ozarks. One night, a man lingered inside the store until the other customers cleared out, then approached the counter with a gun stretched in Kolli's direction. The gunman swiped the cash in the drawer and left Kolli frozen. He felt helpless to stop it.
Eight months later, he would be held at gunpoint again.
He couldn't have known it then, but Kolli would go on to start a business focused on mitigating theft.
His startup MokSa.ai, which aims to help businesses curb theft and employee fraud using artificial intelligence-enabled security cameras, launched from stealth less than a year ago. But it already has a raft of paying customers, which is rare among startups of its size. According to Kolli, its real-time monitoring system is in use at over 70 gas stations, liquor stores, and bodegas nationwide and is generating $20,000 a month in recurring revenue.
Now, this potent cocktail of machine learning and profits has persuaded investors to provide it with pre-seed funding. MokSa.ai tells Business Insider it closed $1.5 million in a March round led by Array Ventures, with participation from Jay Farner, the former chief executive of Quicken Loans, and The Fund Midwest.
The company uses computer vision to automate away some parts of surveillance still done by humans. It makes software for off-the-shelf security cameras that detect suspicious activity — think, a person stuffs a six-pack of beer in their pants or a cashier gives away merchandise to a friend — and sends a real-time notification to the customer dashboard. The alert includes a clip of the incident, which saves the business owner time they might spend playing back hours of camera footage.
Kolli said the company initially tried to use open-source, pre-trained artificial intelligence models but found that some of these models showed biases representative of the datasets they were trained on. It then tried fine-tuning those models, but Kolli said their performance fell apart. Eventually, it landed on using general-use models that continuously learn as new footage comes in.
The company also pays college interns in India to watch footage for suspicious activity and annotate it — a process called data labeling.
"As we sit and speak, there is new data coming in and these models are getting better," Kolli said.
According to Shruti Gandhi, the founder and sole general partner at Array Ventures, part of MokSa.ai's magic is its customized models for this use case.
"The data sets they have gathered in the last two years of being a services company allowed them to deploy trained models in customers from their first week of onboarding," she said. "That is not possible today with a lot of business intelligence companies because most think they're going to get better once they have customer data. So the reason some companies will win over others in the same category is their access to unique, differentiated data."
Given that MokSa.ai's software works with a range of internet-connected security cameras, it's positioned itself as the Android of the surveillance market. The company will also provide a business with the cameras it needs in exchange for signing a multiyear contract. It charges a monthly subscription that scales based on the number of cameras in use.
Customers can pay extra to have an operator call them when the system detects an event, and that's as far as MokSa.ai's responsibility goes, Kolli said. It won't contact the police.
"We are offering surveillance audit platform," he said. "So it's up to the customer to understand what they want to do with that information we provide."
The Android of the surveillance market
Before MokSa.ai, Kolli worked as a quality manager at a company producing parts for high-speed rails. He got the itch to build this company after speaking to a friend who owned the gas station and liquor store where he worked in college. The friend had found out an employee had stolen thousands of dollars worth of merchandise and cash and he asked Kolli how technology might be able to catch bad actors. The idea planted the seed for what would become MokSa.ai.
Later, Kolli enrolled in the startup accelerator Techstars Detroit, where one of his mentors introduced him to his future investor.
Kolli told Gandhi the MokSa.ai story over the phone. He didn't have a pitch deck to show her or any other investor pledges to move her to act. But, the Array Ventures founder bought into the round before the call ended. The next day, Kolli had a term sheet.
The investment was a "no-brainer," Gandhi said.
The company is now gearing up for growth. MokSa.ai just signed a contract with Royal Ozarks, a large commercial real-estate developer in the South, to deploy its system across 150 store locations. Kolli said the partnership will bring its $240,000 annual recurring revenue to over $1 million this year.
The new funding will allow the company to grow its sales and customer support teams as part of this expansion. It's also working to release a mobile app that will allow business owners to view their notifications on the go.
In its bid to create a smarter surveillance system, MokSa.ai faces competition from startups like Verkada, a maker of building security tools that's raised over $360 million in funding, and Rhombus, a newer entrant that just announced $26 million in capital. While these two tout their abilities to detect suspicious activity and send alerts, they both require customers to use their cameras and sensors.
That could be the wedge that gives MokSa.ai a chance despite being much younger than its competitors.
Gandhi summed it up, "What they're saying is keep your camera, we will figure out the best way to use this camera."