Entrepreneur and serial businessman Ritu Srivastava was stunned at how her daughter, now in high school, was navigating online education. This led her to question the level of attentiveness and involvement, of the students who take these online classes.
” Are they studying?” she thought.
“And that’s where the whole conversation began,” Ritu tells Economic Pitch “When things happen within your life, it can give you an idea of a possible issue.”
Ritu has decided to tackle this issue by creating an artificial intelligence that detects emotion (AI) technology that can map emotions during any interaction, whether remote or digital. She collaborated along with Vishal Soni, her husband along with co-founder Yogesh Sachdeva to create Lightbulb.AI.
The company was founded in 2020. company is the third venture into entrepreneurship, after having created and then exited Obino which was a health-related digital platform and weight loss coaching company.
What is the process behind Lightbulb.ai function?
Lightbulb.ai makes use of the machine-learning (ML) methods to recognize the participants’ emotions during video conferencing , or when they are watching content. These data are transmitted in real-time to users to gauge the level of involvement and engagement of the viewers and participants in real-time.
The startup from Mumbai, India, examines faces of people, and then maps the emotions – such as anger, joy, sadnessand more the individual is showing. The basic unit of analysis of Lightbulb is a facial frame which it records by taking pictures and analyzing faces in those images.
In an event with more than one participant hosts or presenters will receive instant notifications with fixed intervals of paying interest. The platform needs participants to grant consent to use their camera to detect facial expressions.
The company claims its accuracy is between 70 and 85 percent across various emotions based on various variables like clarity and resolution of the image, as well as the lighting conditions and so on.
“Our most important metric is our attention or engagement algorithm, which takes a variety of factors into account . Our accuracy ranges from 80-90%.,” Lightbulb.ai says.
The company focuses on three verticals namely remote education, research for consumers as well as sales enablement. It is a horizontal technology platform. Lightbulb.ai states that its platform technology used in these three industries remains the same , but is able to be used in multiple situations.
Its current base dataset has 3.1 million people. The product is offered as a Software-as-a-service (SaaS) product, which is charged on the basis of access and data volume. It starts at $50 per month for each user and increases to $75 per month for users for businesses. The average size of a ticket typically is between $1,500 and $2,000 per month, and is focused on mid-sized companies.
The moment when the lightbulb goes off
While the idea behind Lightbulb.ai was born out of an interest in student, its co-founders realized the technology had many applications in various verticals like market surveys or retail, as well as education.
They began to work on the idea around mid-2020.
The co-founders of Obino had purchased Obino at the end of 2017 the US-based health firm Roundglass Partners. Although Ritu and Yogesh had a purchase clause that allowed them to collaborate in partnership with Roundglass until 2021, in order to aid in the development of a bigger healthcare-related service, Vishal didn’t.
Through Vishal, Lightbulb.ai got a jump on the original product.
“This type that of software is made up of ML which means that you must gather, and train lots of data. It is time-consuming,” says Ritu.
The company has hired two developers, who initially worked on the development of training models as well as a team of data entry workers. Then, in 2021 Ritu and Yogesh were hired full-time.
Initially, the startup obtained data from non-copyrighted sources that had content that was sourced from across different geographies ethnicities, age groups, and genders.
Screenshots taken from video clips were separated into individual frames. Then, the team used annotation labelling, an approach where data is labeled in order so that objects can be recognized by computers. For Lightbulb this meant labelling faces with a certain emotion.
The process is time-consuming even with tools that are available and is followed by a human confirmation of a small sample from the information. Lightbulb.ai has developed its own algorithm for emotion recognition which the data then goes through.
The technology, built upon over 90 facial landmarks, has four patents that are in the pipeline.
The company introduced its first acceptable product, which was launched in the month of September. This validates its product and its market theories. It had approximately 1.2 million users comprising 30% Caucasian and 35% of them comprised South Asian and Indian faces as well as around 35% of them were Latin American, Native American, African American faces.
It began testing the product with a small number of customers before presenting the results to investors for the first round of capital that it announced in the last month. The company was able to bootstrap using its savings for the first year and will continue until July 2022. Lightbulb.ai was able to raise $1.5 million through a pre-seed financing round that was that was led by Chiratae Ventures & 9Unicorns leading the round. Anthill Ventures and VideoVerse, an AI-powered video editing company were also part of the round.
The startup claims it will use the funds to improve the database and further improve its algorithmic ML. It also plans to expand its product and team to different industries.
At present, the company has 11 employees in the team.
“The motive behind this type of financing, as this type of technology is costly to develop and the resources you require are costly to acquire, isn’t sufficient to allow us to operate the company without it but that we would have been at risk if we attempted to operate it from our personal pockets,” Ritu says.
In the early stages of the product’s development, Lightbulb.ai already has three or four paying clients such as Skooc, Harappa Education, Edumocion (a Colombia-based platform for edtech).
The company’s focus is to the market in international markets for its product, which is where there is greater demand.
In this case, Ritu as well as Yogesh’s involvement with Roundglass Partners helps.
“We actually came to a deeper appreciation of the what takes to create business-to-business SaaS (products) to serve the world market, and what the obstacles are and the business strategies that go with the issue,” says Ritu.
While AI-based technology for emotion seems to be a part of science fiction and not real life, startups all over the globe have devised instruments to monitor the users’ emotional responses.
Lightbulb’s product is similar to competitors such as Entropik Tech in a global emotional market that is predicted to grow to $43.3 billion by 2027 at CAGR (compound annual growth rate) of 12.9 percent over the forecast timeframe as per the study from MarketsandMarkets.