An Engineering alumni graduating from IIT Kanpur in 1992, Manish Singhal has had extensive experience of over 24 years in developing hardware and software IP oriented product companies, early stage investing, valuations, deal structuring, and strategy advisory across different sectors. He is primarily accredited to Co-founding of LetsVenture.com, the leading marketplace for startups and early stage investors around the country. He also introduced the initiative called Women Entrepreneur Quest, an annual event which aims at recognizing and rewarding women entrepreneurs in India.
Manish actively participates and coaches the startup ecosystem in the country, acting simultaneously as an advisor and an angel investor. He has also conducted several workshops across different parts of India, guiding both entrepreneurs and investors about how to raise and invest funds respectively.
A sports enthusiast and a published wildlife photographer in his other life, Manish Singal along with Umakant Soni and other key investors established an AI-dedicated fund, called pi Ventures to encourage startups in India who delve in AI, Machine Learning, and IoT. In an attempt to understand the value proposition created by pi Ventures and to paint a more vivid picture about the AI landscape in India, Analytics India Magazine interacted exclusively with Manish Singhal. We present to you the detailed interview as follows.
Analytics India Magazine: While setting up a fund to help AI startups in India, what were the factors that made you invest significantly towards Artificial Intelligence arena? What are the drivers towards promoting IoT and Machine Learning in the country?
Manish Singhal: Last few decades, lot of money and research has gone into building core platform AI, now dominated by big corporations. Now, as these platforms (Tensorflow, Azure Machine Learning, Watson, etc.) are available to use and leverage, the trend is changing. Startups are using some of these platforms as building blocks and applying them to solve real world business problems. This is the realm of Applied AI. We believe that these startups are going to leapfrog and become new category leaders. This is what excites us as investors.
[quote]We are seeing disruptive product work being done across sectors using ML and IoT in India. Healthcare is leading the wave. Other exciting sectors are logistics, fintech, enterprise and retail.[/quote]
India is at an interesting intersection of three factors coming together driving the growth in the field – availability of data, availability of data scientists and willingness of businesses to buy from startups.
AIM: Ten3T is the first company to receive funding from pi Ventures initiative. Please shed light on this statement, talking about the reason to choose a healthcare analytics company as one of the first to receive funding out of this initiative.
MS: We are very excited to partner with ten3T. Dr Sudhir Borgonha and his team have done some path breaking work in bringing out a very usable, compact and medical grade device. We believe that Cicer will make ECG and cardiac health monitoring easy, intelligent and more accessible for all, across India and globally. At pi Ventures, we are looking to support disruptive product companies in the Applied AI, ML and IoT space and ten3T presents a very compelling case.
Healthcare is emerging to be a big area of early focus for us. We are seeing good applications of latest in AI & ML in solving relevant problems in India, specially as India is massively underserved in health care arena, with a doctor to patient ratio of 1:1681 , whereas in USA, it is 1: 400 . Use of AI can increase accessibility of healthcare for an average Indian.
AIM: Are there more companies to receive funding from pi Ventures at the moment? What’s the strategy towards identifying the potential players or startups in the arena that deserve to be funded?
MS: We have made two more commitments. We will be announcing them soon. Apart from the regular factors like a rockstar team, large addressable market, we also look at their Intellectual Property and Data Strategy very carefully.
AIM: Please talk about #chAI session. How is it helping bring collaboration in the AI arena?
MS: #chAI is an initiative to build a stronger entrepreneur community around AI in India. We typically call a practitioner in the space to talk about how they have applied AI to their business problem.
[quote]This interactive talk format helps other founders to learn from each other and create an environment of collaboration across technology as well as talent.[/quote]
AIM: Study shows India to be at the forefront of the change to a AI-based world. Would you elaborate upon this statement, describing the AI scenario in the country?
MS: As per a latest study by Zinnov, India is the third largest AI cluster in the world, just marginally behind UK. We believe three factors are coming together for India for this revolution to happen
a) Availability of Data
b) Availability of Data Scientist and Talent
c) Willingness of businesses to buy from startups
AIM: Can you enlist few challenges in the country that hinder the growth of Artificial Intelligence? What strategies should startups incorporate to solve these impediments?
MS: India still needs to work across the whole spectrum to really make full use of the AI wave. Some of these are,
1. Capacity building in the Area of AI with participation from Corporates, Academia and Startups.
2. Availability of Open Data Vaults, that can allow community to experiment and play with data sets to figure out approaches that can work.
[quote]Netflix challenge built the base of AI community in USA. We need similar challenges to be open to community in India.[/quote]
3. Building peer learning networks, as lot of knowledge in AI is not published yet, so learning from peers might be fastest way to keep abreast of lot of developments. So that is where meetups like #chAI help a lot.
Startups in India do well when they solve a business problem without relying very heavily upon availability of data to begin with. In the process of that solution being used, the data gets slowly accumulated and advanced algorithms can be built over a period of time. We call it Level Two AI startup in our definition. One can graduate to Level Three when you have enough data and good enough algorithms to bring out genuine insights for human action, which can pave the way for autonomous action in future.
AIM: What are the trends in the arena associated with growth of AI over the years?
MS: AI has been in “Winters” for decades. Starting from 1956, when it was formally started in a conference in Dartmouth college; for a long time, even after lot of research into the area, the results were far from being practicable, for almost 60 years. This started to change in 2005, when Prof Geoffery Hinton, realized that neural nets, trained with the availability of faster compute power at much lower cost, made unsupervised learning possible. Unsupervised learning, is the closest to the way, we human beings learn in real life. Thus deep learning was born. Today deep learning is powering, not only the way you get guided by google maps, but also in speech recognition, face recognition and in extracting meaning from text offered by companies such as Facebook or Google.
“Today Facebook can recognize faces (DeepFace) with up to 97.35% accuracy, almost as close to a human”
As such, recent growth in AI has been fuelled by dramatic rise in computational power and data bandwidth in recent times. That has led to many applications using the following main building blocks
1. Computer Vision becoming more Real
2. Sensors becoming more ubiquitous
3. Natural Language Processing becoming more Natural
Now we will see companies that will start to leverage these leaps to solve business problems across sectors.
AIM: Would you leave a piece of advice for all the promising startups and aspiring entrepreneurs who are completely invested towards developing the AI scenario in India?
MS: Focus on the business problem you are solving. That would allow you to evolve a sustainable model for your AI startup. AI is a means to solve a problem, not the end!
Try deep learning using MATLAB