How a bunch of Indian startups have taken an early lead in chatbots

At 11 am on a Wednesday in April, an anxious Rahul Gupta logged into his mobile chat from Amravati, Maharashtra. The first person to ping the 21-year-old was his coach. Gupta, who is preparing for the Common Admission Test, knew why: he had underperformed in a mock test.

“I saw that you just completed an exercise in permutations and combination,” the coach texted. “You got four questions in the exercise,” the coach continued on chat, “and on one question you spent less than 10 seconds. Did you lose your focus or were you over-confident?” Gupta knew there was no point bluffing. “I guessed the answer. I did not understand the concept.”

“You can revise the concept here (the coach gave a link) and you can also solve the exercise once you are ready,” the coach texted back. “Thanks. Will do so,” Gupta replied.

While Gupta might have been relieved that he was not taken to task by his stern coach at Prepathon, an online test prep startup founded in September 2015, what he didn’t know was that the person he was chatting with on the Prepathon app was not his coach but a bot!

Bot—derived from the word robot—is a software that automates the tasks you would usually do on your own. These include booking a cab, making a dinner reservation, fixing an appointment, fetching information, or talking to a customer centre executive. Famous example: Apple’s bot Siri, which answers the queries of iPhone and iPad users.

The most common of all bots is the chatbot, which mimics or simulates conversation using artificial intelligence. This is what the bot at Prepathon was doing. Allwin Agnel, founder of Prepathon, first thought of bots when he realised that the amount of repetitive work done by his teachers or coaches was stupendous.

“We felt we could automate the responses in order to reduce some of this workload,” says Agnel, who founded an educational network for management students and aspirants, PaGaLGuY, in the early 2000s, when he was in his early 20s.

Prepathon, launched in September 2015, raised an undisclosed pre-Series A funding from Blume Ventures this May. Automated responses include answering some basic questions, messaging users about the study topic of the day and sending motivational missives.

Agnel launched the first bot — a topic-of-the-day bot — last December. It messages students the topic they have to complete that day and, if they don’t, the bots will send reminders.

“Bots have reduced human intervention by over 90%,” claims Agnel, who plans to be aggressive in rolling out bots, at least one every month after July.

The bots scheduled for the following month include a feedback bot, which will let users know how they performed on Prepathon; an announcement bot, which makes sure one gets all important updates such as exam deadlines and new course material; a revision bot, which suggests topics that users need to revise based on a spaced-repetition algorithm; and a motivation bot, which will help lift the spirits of students by offering easier goals and showing how their peers are doing on Prepathon.






“Each of these bots will use varying degrees of machine learning and artificial intelligence to keep getting better at what they do,” says Agnel. Bots at Prepathon answer over 10,000 questions a day, he adds.

Also Read: Bots will gradually invade every aspect of our lives and transform the way we use mobile devices

The Bot Invasion
Agnel’s zeal coincides with frenetic activity around bots in the West. Global tech giants Google, Facebook, Microsoft and IBM along with SpaceX founder Elon Musk have all made various bets on conversation-capable products, intelligent messaging apps, partnerships between people and computer, and artificial intelligent research (see Global Tech Giants & Brands Bet Big on Bots and AI). “Bots are the new app,” declared Microsoft CEO Satya Nadella at a conference in San Francisco in March.



“That’s the world you’re going to get to see in the years to come.” Facebook CEO Mark Zuckerberg then weighed in. “No one wants to have to install a new app for every business or service that they want to interact with,” he said at a conference in April when he announced the opening of Messenger to bots. The reasons for the scramble to take a lead in the bot revolution are not hard to fathom:

Millennials — those born between 1980 and 2000 — are texting more and talking less. And millennials matter because they will account for 40% of all consumers in the US by 2020, as projected by Goldman Sachs. What’s more, text messages outranked phone calls as the dominant form of communication among millennials in the US, according to a 2014 Gallup poll.

In the first quarter of 2015, the four biggest messaging apps surpassed the four biggest social networks in terms of monthly active users.





The writing is on the wall: users are fast getting afflicted with app fatigue. Though thousands of apps are launched every month, only a handful of them gain consumer traction. American consumers spent 84% of their time on just five apps last year, according to Forrester Research.







Early Movers in India
Indian consumers are following in the footsteps of their American counterparts. “Bots will move beyond chat interfaces,” says Agnel of Prepathon. Agnel is not the only Indian entrepreneur who is looking for an early mover advantage in bot adoption. Manish Taneja, cofounder and CEO of Purplle, an online marketplace for beauty products and services, too wants a slice of a future littered with chatty products.

A few months back, Taneja realised that a lot of queries answered by his customer support executives were repetitive in nature — on order status and delivery timelines. Taneja also noticed that user queries on product discovery were on top-rated shampoos for dandruff, nail art and long-lasting makeup for oily skin. It was then that he decided to develop a chatbot. “We have done a beta launch of the chatbot for our most loyal customers and the response has been amazing,” claims Taneja.



The bots have reduced by at least two-thirds the load of the customer support staff by making the information available on demand and by automatically executing certain tasks. Currently, the bot helps around 2,000 people everyday, responding to queries on product discovery, he adds. “Chatbots are here to stay.”

Will Bots Beat Apps?
Bots, aver its makers, have an edge over apps. For one, they don’t need to be downloaded as they reside within the messaging apps. So they can be used instantly.” For another, they are easy to build and upgrade, and faster and cheaper than apps. They don’t choke your mobile by taking a lot of space like apps do,” Taneja explains. Sachin Jaiswal, cofounder of Niki, a shopping assistant chatbot, not only wanted to tackle the “app choking” problem but also wanted to get more Indians to transact online.

While more than 250 million Indians are on the internet, only 20 million transact online. It didn’t take long for Jaiswal to figure out where the rest were: on messaging apps, roughly 150 million of them. To build a unified transactional interface, Jaiswal realised that chat would be a great way to start. “So Niki was born.”

The startup was founded in April last year and the product was launched in October. It counts Ratan Tata and Unilazer Ventures among its investors. Niki has tied up with Ola, Uber, Burger King, Paytm, Mobikwik and JustRechargeIt to offer services such as bill payments, cab bookings, recharge, home services and food ordering.




While the startup is planning to venture into hotel and bus bookings this month, over the next quarter it has set an ambitious target to provide 25 additional services such as ordering grocery, buying movie tickets, booking flights and paying for healthcare. “Niki aspires to become an ideal shopping assistant for one and all,” says Jaiswal.

He sees potential for Niki for three reasons: one, users in metros are juggling multiple apps, too many distributed options and non-interactive interfaces. Add to this bad connectivity in Tier II cities and phones with low memory. “Niki is a lightweight messaging platform that works seamlessly even on 2G,” he says, adding that users over 35 years have a big learning curve when it comes to using apps, ordering on them and making payments. “With its simple chat interface, Niki can solve these problems,” he claims.

While Jaiswal might be content with one chatbot, others want multiple bots. Aarti Gill is one of them. The cofounder of FitCircle, a chat-based health and fitness platform, boasts 32,000 bot users. Gill, who launched her first bot in November last year, now has five of them, each performing a different task. If one helps users with personalised workouts in yoga and for weight loss, another focuses on diet and nutrition advice, with expert intervention wherever required.

As FitCircle readies to launch nutrition products under the label O’Ziva, which will include meal replacement shakes and protein supplements, bots will focus more on consumer education and helping users find the right product. According to Gill, chatbots present an enormous, new opportunity. With bots, brands have the power to have conversations with consumers through a combination of spoken or typed phrases, which gives an overall personalised experience as opposed to apps that force users to choose from available options from a drop-down menu. “The fact that bots let you instantly interact with the world around you is what is exciting all the players globally,” she says.

The opportunity might be massive but the challenges are equally daunting. For the time being, the biggest one is minimal awareness about bots among people. “A lot of people either don’t know about bots or have very different expectations from them,” says Sumedh Chaudhry, cofounder of Gurgaon-based media startup NewsBytes. Though getting a user activated on the bot is at least five times simpler than asking her to download the app, the issue is awareness, rues Chaudhry, who claims to have over 8,000 bot users.

The startup got $200,000 seed funding last April and launched its first bot in December 2015 on Slack, a business-focused messaging service. This April, it launched its bot on Facebook Messenger. “If you educate people, acceptance and usage increases exponentially,” he says. Chatbots can at times go terribly wrong. Remember Tay from Microsoft, which turned racist and abusive in March.

The bot, which was supposed to mimic the verbal tics of a 19-year-old girl on Twitter, Kik and GroupMe, was turned off in less than 24 hours as it started promoting Nazi ideology and harassing other Twitter users. Chaudhry says he has factored in the possibility of users playing pranks with bots. He thinks the only way to solve the problem is through extensive training of bots.

“Think of bot like a human baby who has a lot of curiosity and appetite to learn,” he says. The baby can pick up the wrong habits or she can pick up all the right things and become the smartest kid on the block. It depends on the kind of exposure the baby has, says Chaudhry.

The bot has to constantly evolve, every day, every hour. “It has to learn a lot of things, and then unlearn a lot more,” he says. Jaiswal of Niki too tries to prevent the Tay kind of abuse. “Being a conversational commerce platform, we have to consider these possibilities,” he says. If someone is trying to pull a prank or use abusive language, we have a procedure to block the user for some time after a set of warnings, adds Jaiswal.

Another problem bots might face is irrelevant questions from consumers. What if, asks Taneja of Purplle, the bot is not able to answer such queries satisfactorily as it is not programmed to do so? For instance, a user can ask: Will it rain today? Now the bot won’t be able to answer this. “We have taken a conscious call to train our bot conservatively,” says Taneja. It is better to err on this side than to have the bot reply distastefully upon manipulation. A standard approach is for the bot to ask the user to rephrase the question, while politely reminding them about the scope, he adds.

Technology is a double-edged sword, says Taneja, and one should be equipped to handle the flip side as well. He points to the evolution of Siri into a sarcastic, sassy virtual assistant. A year ago, iPhone users realised how soul-crushing Siri could get when it replied to what is 0 divided by 0: “Imagine that you have zero cookies and you split them evenly among zero friends. How many cookies does each person get? See? It doesn’t make sense. And Cookie Monster is sad that there are no cookies, and you are sad that you have no friends.