The best home-grown AI start-ups6 min read . Updated: 24 Sep 2016, 12:55 AM IST
Digital assistants you can chat with, solutions that learn your taste in clothesthey're now being made in India
The word on the streets of India’s start-up world is that any mention of artificial intelligence (AI) makes venture capitalists drool. From Sequoia India to Tata Sons Ltd’s chairman emeritus Ratan Tata, reputed investors are adding companies working on AI to their portfolio. This is only to be expected, given that the entire tech world can’t stop talking about how AI, machine learning and neural networks will completely change the world over the next few decades. But it’s also a daunting task for Indian companies, given that all the big global players are pouring massive resources into the space. There are several Indian companies that have taken AI out of the lab and found real-world applications for it.
The screen you can talk to
If you walk into the Nepean Sea Road branch of RBL Bank Ltd (formerly Ratnakar Bank Ltd) in Mumbai next week and swipe your card, you will see yourself on a large screen that will enable various banking tasks, such as opening a new account or applying for a loan. You can interact with the screen using gestures, swiping your hands right and left to click on buttons and see further options, and you can also talk to the screen.
As you view the options, the screen will study your face, trying to gauge whether you’re engaged, interested or bored. If it believes you’re interested in car loan options, it will suggest schemes on these.
All this is powered by Fluid AI, a Mumbai-based start-up founded three years ago by brothers Abhinav and Raghav Aggarwal. Their gesture and speech recognition solution not only lets companies provide customers with novel and intuitive experiences, it also provides the enterprises with better data analytics.
As you interact with Fluid AI’s solution on the screen, it studies you and your preferences. It then crunches all this data and presents the enterprise with the analysis.
Apart from RBL Bank, Fluid AI has also built a solution for Axis Bank, which it will launch sometime next month. Auto companies such as Rolls-Royce, Hero and Toyota, as well as mobile service provider Vodafone, are among its customers.
Fluid AI is also helping enterprises use AI to make real-time decisions based on data. Their AI can trade in the stock market, says Raghav. It learns when to buy and sell a stock based on past data and can then make the actual decisions without human intervention. Similarly, if a mobile phone company wants to know when its customers are thinking of changing their service provider, Fluid AI can learn customer behaviour by analysing data—dropped calls, complaints, poor network areas etc.—and reach a decision on making a particular customer an offer or escalating his complaint at the crucial moment.
If you search for Niki Bot on the Facebook Messenger app, you’ll be introduced to a virtual assistant that can help you pay your electricity, phone or cable bills or book you a cab. Type in “pay my phone bill" and it will ask for your number, service provider and bill amount. Then, if you have a Paytm account, you can pay your bill right there on Messenger, without being redirected to another app or site.
Niki Bot is the creation of Niki.ai, an year-old Bengaluru-based start-up that has raised funding from, among others, Ratan Tata and UTV group founder Ronnie Screwvala. Their aim is to allow you to conduct various transactions online—paying bills, booking transport, ordering food, getting chores done—on one platform, while talking to a digital assistant.
Apart from their Facebook bot, they also have their own app, Niki. On this, you can do everything you can on the Facebook bot, plus book a bus, order a burger from Burger King and get your laundry done. What makes the app “intelligent" is that it will attempt to have a conversation with you, while an app like Zomato would make you go through menus and select things yourself. If you want to order a burger on Niki, for example, you can say “I like chicken", and it will tell you it has a grill chicken burger combo meal available. When it asks you when you want your laundry picked up, you can say “First thing in the morning", and it will understand what you mean.
MagicX, a Bengaluru-based start-up that raised funding this year from investors such as Kris Gopalakrishnan, a co-founder and former chief executive officer of Infosys, does almost exactly what Niki does. Its app and Facebook bot are both called MagicX, and the app can communicate in English, Hindi and Kannada—eight more languages are expected to be added soon. The company has the advantage of having previously operated a chat application called Magic Tiger, which had people answering requests over messaging, so it has gained some insight into what customers want from a digital assistant.
Neither Niki.ai nor MagicX are doing anything revolutionary. There are plenty of digital assistants powered by AI being developed around the world, not the least Facebook’s own M, which lets people in the US perform all sorts of tasks, from buying someone a birthday gift to making dinner reservations. However, they can stay ahead of the competition by quickly tying up with as many service providers as possible. So if a company such as Facebook does want a digital assistant that could perform tasks for people in India too, it would be easier for it to simply acquire or ally with Niki or MagicX rather than approach every restaurant, airline and mobile service provider for tie-ups.
The E-commerce enablers
Right now, e-commerce offers convenience and lower prices than physical stores. But the next step is to completely revolutionize shopping by providing truly personalized experiences. What online shopping sites and apps mainly rely on now are tags keyed in by people. If you search for a sari that an app has tagged as Banarasi, you will see ads for Banarasi saris the next time you open it. But the app is not truly learning your taste in clothes.
This is where Mad Street Den comes in. A two-year-old Chennai-based company, Mad Street Den is specifically targeting online fashion retailers and, through its tool Vue.ai, helps them learn more about their customers and make the experience of shopping more intuitive. Using Mad Street Den’s services, retailers can also allow people to search for garments through images rather than text.
So, rather than typing in “green sari with sequins suitable for casual and formal functions", you can just take a photo of a sari that fits that description, and the online shopping app, using Mad Street Den’s AI, will show you similar saris. All the while, the app is learning about your specific tastes and will suggest products based on its assessment.
Mad Street Den counts online fashion retailers Craftsvilla and Voonik among its clients. Slightly worryingly for the company, though, Flipkart recently discontinued its image search feature, which was powered by Singapore-based ViSenze, citing minimal usage.
Mumbai-based Arya.ai is also hoping to enable companies to use AI, but rather than building tools for them, it has built a programming tool called Braid, released earlier this month, that other companies can use to build neural networks. Companies can then use those neural networks to do everything from automatizing their processes to learning how to recognize photographs.
For companies such as Mad Street Den and Arya.ai, the competition will be stiff, as some of the biggest tech companies in the world are researching image recognition and analytics. A recent Evans Data Corp. report says that about 60% of software developers in India have been experimenting with AI and machine learning. So we can expect plenty of competition in the space in the next few years.
Jason Fernandes, founder of tech start-up SmartKlock, contributed to this story.