The journey of wesense.ai—as the CEO Sankara Srinivasan explains—began with another startup called Realtycomforts.com. The three co-founders—Srinivasan, Soumya R. Mohapatra and Alok Mishra—were already a hit with the investor community. Post-acquisition of the startup and ending the lock in period, they wanted to begin another venture. And this time the aim was to help non-digital business grow by providing them with actionable insights.
Thus began—wesense.ai. Non-digital businesses have only transactional data to go on with, therefore the necessary information on customer behavior was lacking. But Srinivasan says, “The camera and computing is there, only intelligent software is missing. We can combine together and give actionable insights.”
The startup created a CCTV camera brand agnostic platform, thereby making it compatible with analog or digital and feasible with CPU, rather than a GPU. Given their work history in corporates and earlier entrepreneurship experience the team was able to reach out to multiple industries but zeroed down to retail. The sector would gain maximum benefit from the platform and “they are very much dependent on the customer behavior data for their revenue,” Srinivasan says. The team, he explains, got the best of the customers including two of the Fortune500 and the likes of Dell, Xiaomi, Tata Group, Mahindra Group, etc.
What is the road to the coveted actionable insights?
The CCTV camera is connected to the DVR or NVR that stores the video. The team created an app store with 10 apps—operable on Windows, Linux. Now, any customer or store owner can select an app basis the need of the business. For instance, you would like to know customers carrying an iphone or when people visit the store in a group what’s their shopping pattern or how long people wait in the billing counter, etc.
The app installed in the computer, connected through the LAN to stream the video. The app will fetch the video, crunch it and give the output depending upon the query. “It fetches that particular cam according to the query and starts showing the necessary insights. These actionable insights then go to the manager. He can see the problem area and make decisions basis that–more people are waiting at the billing counter, might open another one,” says Srinivasan. “The point is to get absolute data, compile it with a business logic.”
Wesense.ai’s customers form the high engagement retail demographic. Therefore, these retail chains have to have high engagement which leads to higher sales. “Our algorithm can identify the employee by multiple parameters using face recognition, uniform color, logos and id when a person enters and for the whole day, we can identify the person with the same id. We track how long the person stays inside the store.
We identify the engagement time between the employee and the customer. We also identify the aggregate level of engagement happening in these stores and at the individual level and link that with the sales,” says Srinivasan.
Srinivasan and team could do all the tech but would still need human intervention to identify and understand certain measures. For instance, a store, Srinivasan explains had people standing for a long time, which led to them exiting the place without making a sale or transaction. A simple action of providing more chairs, which was pointed out post understanding from the intelligence, led to improved conversion rate. “It is about resonating with people behavior. You still need a tech and human intervention to cull out actionable insights,” he says.
Additionally, the product provides privacy by design. “We don’t identify individual people, but give aggregate data. Second, the processed data is masked to prevent it from being tracked. We don’t store any personal data, everything is processed in the customer’s own systems. Only the meta data comes to our cloud,” he explains.
What differentiates them from the rest of the similar solutions in the market is the solution’s ability to work across any CCTV camera brand and in any environment, says Srinivasan. “We don’t create capex,” he says.
Second, he says, the time to go live is less than 10 minutes. “We have more than 5 million data points, process at the edge level and give high accuracy,” he says. “We are result-oriented. Rather than just giving data we go beyond and see how we can add value and deliver result.”
The state of Indian Deeptech ecosystem:
Srinivasan believes that though consumer tech has flourished with internet and smartphone penetration, Deeptech has a long way to go as it is primarily on an enterprise or B2B level. “But Indian businesses aren’t comfortable using the new age tech. It is then a problem is to grow certain revenue in a quick period if you depend on the Indian customers,” says Srinivasan.
The path forward should be to build from India and sell globally, though there might be challenges pertaining to capital and travel and the architecture compatibility, he explains.
“Unless the enterprises or large conglomerates start using the startup products in India, it will be difficult for the deep tech companies to create a certain revenue,” he says.
Srinivasan suggests that the government can help by collaborating with the deeptech startups in the government projects. That can help get the base level knowledge, which could snowball to better companies and markets. “The government has to include AI into the mainstream and as many startups as possible in their program. Then we can see deep tech startups come from India, otherwise only few will survive the turmoil and most will perish,” he concludes.