IVA taps new market with surveillance software

Necessity is the mother of invention, and the team behind the Trivandrum-based startup International Virtual Assistance (IVA) would wholeheartedly agree with this age-old saying.

IVA’s core expertise lies in aerospace and data analysis, with the team playing a key role in developing India’s indigenous jet engine. They have been involved in the Gas Turbine Research Establishment’s (GTRE) work for the past couple of years, as part of the Kaveri engine project of the Defence Research and Development Organisation (DRDO). The engine is now being manufactured by Godrej.

Despite this commendable achievement, IVA faced a challenge familiar to firms in the Indian deep-tech sector: a funding crunch.

These firms often spend years researching to create indigenous technology that rivals established global players. However, they struggle to raise funds to stay afloat.

Securing major defence or aerospace orders involves extensive vetting processes and can take a long time. Even after landing multi-million-dollar orders, many firms find it difficult to raise working capital, as few funding firms cater to this need, and grants for the deep-tech sector are lmited.

IVA admits that the work they are doing for the defence sector is a privilege they cherish, and it helped upgrade the skills of their engineers to a world-class level. However, they were spending about 70 lakh rupees annually while working on the GTRE projects, but their revenue was not keeping pace.

IVA continued to collect accolades from different events, the latest being adjudged the best research company in AI during the fourth Global Artificial Intelligence Summit and Awards in 2024.

It also won recognition from the All India Council for Robotics and Automation (AICRA) for its partnership on defence-related projects to develop a virtual strain gauge sensor for aero-engine testing.

Still, the company needed to have a steady cash flow. That led it to develop a surveillance tool called NxtEye, which uses artificial intelligence to analyse security camera feeds, send instant warnings when a breach occurs, and even predict potential issues.

IVA says their tool can incorporate feeds from any camera, including analogue ones, without requiring additional sensors.

“We see this as a vertical that can bring in good revenue, as the system can be used by customers who have multiple cameras and struggle to keep track of all feeds simultaneously,” says Aji Soman, Chief Operating Officer of the firm.

NxtEye has already impressed significant clients like Fortune Group and Malabar Gold, who have entered into deals with IVA. Talks with more companies are going on.

“But DataCivet remains our core work. It was created for data processing, storing, and building. Its advantage is the ease with which it can be configured. You don’t have to be a machine learning expert to build a model using DataCivet,” says Varun VV, the Chief Technical Officer of IVA.

IVA’s journey in that direction started during the 2022 IDEX competition when it became the only Indian company selected to pitch its skills in the auto machine learning model (AIML) category.

They made it to the final round, competing with firms from the US and other countries. However, like many Indian deep-tech firms, they lacked customers and revenue, which proved to be their undoing in the competition.

Learning from that experience, IVA approached DRDO, which granted them a project. This relationship has continued for the last 22 months.

The guidance of chairman Suresh Kumar, former Associate Director at DRDO, and Roy George, Chair of Cyber Physical Systems at Atlanta University, has been invaluable in this work. Both graduated from the College of Engineering in Trivandrum and have enviable track records in their fields.

When IVA approached DRDO, its gas turbine team was facing a difficult challenge. The tests were done in Russia and the engines, with over 1,000 sensors in it were being flown there. The sensors had to withstand very harsh conditions at Mac 1 speed and if a couple of sensors among them fail, the test would be unsuccessful.

The task IVA took up was to see if they could develop a method to address this through the creation of a virtual sensor by collecting data that was available from other sensors. It was a high-precision job but IVA managed to come up with an effective solution.

The work in the defence organisation came with severe challenges as IVA engineers were not allowed to take their phones into the restricted area and the internet also was not accessible there. So, they had to draw up codes without any additional help.

“We had to do everything from our memory. It was hands-on experience in its truest sense. No cut-and-paste business was possible,” says Varun.

As IVA continued to produce results, GTRE’s confidence in them increased. The continued association gave IVA engineers the opportunity to work with data that only a handful of global companies, like renowned fighter jet makers, can access. This enhanced the capability of the team immensely.

During that time the IVA team noticed that there were around 150 security cameras in the complex, making it impossible to monitor the feed from all cameras. As the team had developed the capability to handle large volumes of data, they started working on a model that could manage the feed from a network of security cameras.

Most surveillance cameras save the images onto a hard disk or the cloud. If a problem occurs, the stored feed will be analysed by experts. This process is cumbersome, as scanning could take hours. So IVA developed a system that could perform real-time analysis of the feed from the cameras and send alerts when an anomaly occurs.

“The beauty of NxtEye is that it could be integrated into any existing camera network without tampering with them or adding extra sensors. It could use feeds from cameras of different kinds, including analogue ones,” says Varun.

The system’s versatility means its market potential extends beyond security, and can be applied in areas like healthcare. IVA is now aiming to expand into the Middle East, which it believes is a promising market for NxtEye.

IVA now generates around 1 crore rupees annually, but with 22 employees and work going at full throttle, it needs new revenue streams to keep up the momentum. So a market for NxtEye could be crucial for the firm.

Getting business out of government projects is difficult as it requires processes like tendering and listing on the central government’s Government e-Marketplace (GeM) portal. However, Delhi police have expressed interest in their AI analysis of traffic cameras and predictions about accident black spots.

Aji Soman says work in aerospace will remain the main thrust, and now the company is looking at a much more ambitious project: creating a virtual twin of jet engines that will provide the perfect platform for experts to try new designs.

“While this work goes on, we realise other verticals are needed to generate revenue. NxtEye is the first step. More products will follow,” he says.

Dogfight continues for AI crown

The scramble for the AI top spot continues between OpenAI, Google, and Meta. OpenAI has rolled out its SearchGPT in a bid to chip away at Google's dominance in search. Some initial reactions from users who have access to it are not very impressive. CNBC's Kif Leswing posted on Twitter that some of the search results were inaccurate and not helpful. Meanwhile, Meta has released Llama 3.1, an open-source large language model. As TechXplore says, Mark Zuckerberg is probably the last person you would expect to embrace open source. But that has happened, thanks to the AI greasy pole.

While all this was going on, Google’s DeepMind announced that its specialised AI systems have solved complex math problems, achieving a level equivalent to winning a silver medal at the Math Olympiad. “This is the first time any AI system has been able to achieve medal-level performance,” says Pushmeet Kohli, Vice-President of AI for Science at DeepMind. “This is a key milestone in the journey of building advanced theorem provers.”

Virtual sales engineer gets a nod

DocketAI, the latest venture of former Chief Data Officer of ZoomInfo, Arjun Pillai, has secured backing of 15 million US dollars from investors led by Mayfield and Foundation Capital, reports TechCrunch. Pillai, who co-founded the firm with CTO Anoop Thomas Mathew, says the venture aims to save the time of technical engineers who have to keep answering repetitive questions from different clients. The virtual sales engineer will take care of that now. Pillai tells Foundation Capital that the firm got its first funding within days after it started formulating the idea.

Google offer fails to sway four Wiz men

Tempting offers are difficult to brush aside, especially when each of the co-founders stood to gain 2 billion US dollars. But that is what the co-founders of cybersecurity firm Wiz did. They turned down an offer from Google, which was offering 23 billion US dollars, even though the company’s valuation stood at 12 billion. “While we are flattered by offers we have received, we have chosen to continue on our path to building Wiz,” said Wiz CEO Assaf Rappaport. But CNBC reports that antitrust and investor concerns were the reasons why the startup rejected the deal.

Book your space in advance

Plan your holidays well ahead. That’s advice the folks at TripAdvisor take to heart. So, they’ve put up “Destination Space” for those planning to zoom off to the final frontier. It even has the best spots you should visit and reviews by people who’ve already been there. Sian Proctor, who joined the 2021 SpaceX flight, offers tips on keeping the spaceship clean and the best selfie spots once you’re up there. Hold on, don’t start looking for spacesuits yet – the site doesn’t accept ticket bookings.