The scene begins with an otherwise shy Dr Raj Koothrappali welcoming his guests to a dinner party he’s co-hosting with his new-found friend. He enthusiastically shares details on how his co-host helped narrow down the grocer for the meal ingredients, the perfect wine and playlist for the night. All was fine till his unassuming guests inquired about the co-host only to discover it was none other than Apple’s digital personal assistant, Siri! Dr Koothrappali, the introvert, had found solace in a bot that was supported by artificial intelligence. While that scene belonged to a 2012 episode from the sitcom The Big Bang Theory, the likelihood of something similar playing out in 2017 at your own friend’s dinner party is quite high.
Artificial intelligence-supported bots are the apps of today and start-ups are busy brushing up their machine and deep learning skills to help you find that perfect smart companion for all your needs. To begin with, artificial intelligence (AI) is a branch of computer science that is related to developing machines that are capable of human-like intelligent behaviour. Machine learning is a program created to train the device using troves of available data and deep learning is an advanced form of it that has gained popularity in the last few years.
As per data provided by research firm Tracxn Technologies, the number of AI-based start-ups in India has risen from a measly seven in 2011 to around 300 till April 2017. And it also states that the amount raised by such start-ups has increased from $0.22 million in 2013 to $20.50 million thus far this year. While the core technology remains the same, different start-ups are applying this for multiple purposes by focusing on processing and analysing different forms of data — text, audio, video, images, etc.
Chatbot-based digital assistants are the most common application of deep learning deployed by Indian start-ups. Apart from assisting the user in a format that is easy to interact with, it helps to eliminate the need for other apps. Storage crunch on smartphones is a reality today and users don’t think twice before swiping an app into the bin if it doesn’t serve their purpose.
Bengaluru-based Niki.ai was one of the first few Indian start-ups to develop a chat-based assistant, Niki. “Bots can automate a lot of repetitive tasks especially in the service field by being supplemented with AI. Our ultimate goal is to create a marketplace for e-commerce enabled by chat,” explains Sachin Jaiswal, co-founder, Niki.ai. Launched in 2015, this chatbot manages to strike up to 500,000 conversations a month with its 80,000 active users.
Once you download the app and fill in your basic details, this human-intelligence augmented bot can help you pay your bills, recharge your phone, book a cab or order a meal.
Niki has tied up with 40 merchants such as Burger King, Ola, Uber, OYO Rooms, from whom it earns on an average a commission of 10% for every completed transaction. Having clocked about Rs.6 crore in revenue so far, Niki is the first chatbot to have integrated a mobile wallet within its app — Paytm. “One of the things we liked about Niki’s strategy was its approach to work with big brands directly. And the second reason, it has a simplified chat interface that anybody can use,” says Ankit Shah, VP, Unilazer Ventures, which has invested in the start-up right from its seed funding to its second round of investment in 2016. The other prominent investor includes Tata Sons chairman emeritus Ratan Tata.
As per a Google report titled Mobile App Marketing Insights: Asia dated October 2016, on an average an Indian user uses only nine of the 33 apps installed on his/her phone. And the top criterion for an app to be used daily is that it makes life easier. This crucial insight is the reason Haptik co-founder Aakrit Vaish changed his business model.
Launched in 2013, Haptik fields a million requests a month from its base of active 250,000 users currently across nine categories — travel, food, reminders, shopping, cabs, home services, recharge, movie tickets, and locating nearby services like ATMs, bus-stops and petrol pumps.
“A year after we began operations we found that the requests from customers were less about complaints and more transaction-based like booking a movie ticket, buying a smartphone with a good deal, flight check-in, etc. That’s when we switched from a customer service avatar to a personal assistant model,” he says. The start-up has signed up 22 partners like Flipkart, Snapdeal, Zomato, DineOut, etc. After a $1-million investment from Kalaari Capital in 2014, Haptik received another $11.2 million investment in 2016 from Times Internet. Miten Sampat, VP, corporate development, Times Internet, says, “Haptik is finding an efficient way to perform daily tasks for users. What sets it apart from other similar chatbots is its mix of human and machine intelligence.” Haptik’s 350 employees work across multiple shifts to fulfill requests that the bot hasn’t attempted. Unlike Niki, which claims that all requests are answered by bots, Haptik’s bot handles 40% of customer queries and Vaish is keen to increase that number gradually as the bot learns.
“We will obviously add more categories of services in the future and the bot will also improve with time. We want to make Haptik a habit for the user. The app should be able to perform tasks for you even if you have not opened it yet,” adds Swapan Rajdev, co-founder. This isn’t an unrealistic ambition for companies such as Haptik given how the space for digital personal assistants is heating up. Global giants right from Amazon to Google to Microsoft have not only launched their respective digital assistants but are also constantly improving their efficacy to ensure that it provides more contextually accurate information to a user within seconds in response to voice commands. Newspapers every day proclaim a new winner in the race to AI supremacy with Apple’s Siri getting smarter, Microsoft’s Cortana being sharper and Google Assistant being snappier with the recently launched Pixel phone.
Apart from assisting users in their personal lives, AI is also making its way to the work space. Instead of appointing humans to read through piles of data to pick out insights, start-ups have written a code that does the job for you with a human-like script.
Those cumbersome quarterly reports that you once meticulously prepared over hours can now be prepared by vPhrase Analytics’ Phrazor. “It summarises data in words. All companies have to work with data and as such our product is being used for varied use-cases by companies in multiple verticals — BFSI, FMCG, media, e-commerce and more. We set up and deploy in less than a week,” says Neerav Parekh of the patent pending software that analyses data and represents it in a personalised narrative at a fraction of the cost and time taken otherwise. vPhrase, which received $300,000 from Venture Catalysts last year, counts the likes of HDFC Bank, Accenture, Dun & Bradstreet and a couple of financial institutions in Canada among its clients. “Currently, vPhrase is a leader in the natural language generation space in India. There are many companies who are into reading language but very few are into writing. Globally, there are only three to four other companies in the same space as vPhrase,” points out Apoorv Ranjan Sharma, co-founder, Venture Catalysts.
In India, Wipro’s AI platform, Holmes, provides natural language generation program-based software products that include predictive systems and visualising knowledge. That space is also seeing tough competition with TCS launching its neural automation system for enterprises called Ignio in June 2015.
Making things smarter
In addition to text, start-ups are also developing AI systems that can analyse videos and images in real time. Bengaluru-based Uncanny Vision specialises in the AI field of computer vision which makes camera-enabled devices smarter. “Today the problem is that there are surveillance cameras where data is stored for post-event analysis. But how can one process this data in real time and get actionable insights to prevent or reduce crime?” asks Ranjith Parakkal, co-founder, Uncanny Vision. Launched in 2013, this AI start-up developed a software that trains the machine to analyse live video feed as required — like face recognition, identifying body movements, etc. “We use a few days data to learn what is usual and unusual about a scene. The camera is then programmed to raise an alert when it spots an anomaly,” explains Parakkal. The software is currently deployed at manufacturing units in Japan, US retail companies and even at an Indian private bank’s ATM outlets on a trial basis and costs anywhere between $20 and $50 per camera.
Most of the AI start-ups attribute the spurt in the rise of such companies to the democratisation of deep learning techniques. Developers and universities across the world are making their code open source to have more people involved in improving the system. One such company is Mumbai-based Arya.ai that developed a machine learning language, Braid, and declared it as open source. Started in 2013 with the idea to provide a research assistant to the academic community, Arya soon realised that the lack of understanding about AI’s benefits meant, it was not a monetisable code.
While AI-labelled start-ups are mushrooming every day with a focus on any one application there are also many shutting shop as they couldn’t keep up with the fast evolving space of deep learning. That’s when co-founder Vinay Kumar, decided to move onto developing an AI platform that helped companies build products for their own use. “Our work-bench platform, Vega, is like an operating system for those wanting to build AI platforms. We provide the software on a subscription basis and if needed also provide a consulting service for the same,” explains Kumar.
In addition to the usual business automation, Vega also finds use in the banking and finance sector for financial modelling, portfolio construction, risk prediction, etc. Currently, the company counts some of the top financial institutions such as Zurich-based SwissRe and French bank Societe Generale among its clients. The base product could cost anywhere between $3,000 and $6,000 for a month’s subscription. Arya has also made its way to the global innovators’ list by French&Co this year along with being one of the six Indian start-ups picked by SwissRe for its first insurance focused accelerator. The company raised $750,000 from Venture Nursery and Your Nest Angel last December. And Girish Shivani of Your Nest shares more about what motivated his decision to fund a start-up in an emerging space that didn’t even have a ready product back then, “Unlike other companies which had point-focused AI solutions, Arya is looking at creating a platform to facilitate that process for other enterprises. We liked them because of their different approach to AI, its workbench model can be adapted to build any AI product.”
While investors are gung ho about their AI picks, they are hoping for a more warm reception from Indian companies. “The problem in India is that corporates are not very start-up friendly. They make all these promises at press conferences but the decision-making process is very slow,” says Sharma adding how it took just 15 days for vPhrase to enlist three of Canada’s top banks for a data analytics trial.
While advanced AI may be a new field and enterprises are nimble footed in their adoption, some are already eager to push the boundaries. For instance, Japanese semi-conductor major’s US arm Renesas Electronics is one of Uncanny Vision’s smart clients. Raj Nagarajan, senior product manager, Renesas US, says, “Uncanny currently provides solutions on the Edge, in the future the play will be on cloud computing and everything will eventually move to the cloud.”
Another such AI client, restaurant discovery and reservation start-up, DineOut is a partner with personal assistant provider, Haptik. And DineOut’s co-founder Sachin Jain is already working on making the most of this partnership to improve his margins, “Personalisation is the next big step. For all our partners like Haptik to whom we provide APIs, we could also add valuable insights on each customer which will equip the bot to serve him better.”
Some companies are already racing ahead in the bid to develop personalised AI bots. Hyderabad-based TalentSprint, a youth career accelerator, has developed a digital learning assistant called TIA for its 30,000-strong user base. “It is humanly impossible for one instructor to personalise a lesson plan for a million students. But, TIA can provide personalised feedback on a test and recommend study material for every learner,” explains Santanu Paul, CEO, TalentSprint. And he envisions the bot to do a lot more as it ages with data.
The emergence of AI in India is a reality with budding enterprises and large IT behemoths experimenting with multiple applications catering to diverse sectors. Investors feel that the key lies in securing big clients with renewed gusto in mature markets such as the US and Europe instead of waiting for Indian companies to come around. And these start-ups which already have enlisted a few firms of repute in the global market haven’t taken their eyes off the core AI innovation space as they realise that it’s the only way to stay relevant.