There is a price to be paid to be the first

Srikanth Velamakanni, Co-founder and Group CEO, Fractal Analytics, and winner of the Dataquest Path-breaker of the Year award for 2024, looks back and analyses the dots that led the company from data to AI, from analysis to predictions and from plain AI to Gen AI. Excerpts from a chat with Editor Sunil Rajguru.

As a company that is over 20 years old, it seems like a long journey- especially how the country has gone from the smartphone revolution to the pandemic. Can you share your story from the inception till today?

We started in 2000, during the peak of the Dotcom bubble. The initial idea was not even about being an AI firm but about helping consumers make better decisions on what they buy. Then we took a pivot of helping companies make better decisions. We knew we were good at human behavior, Maths and Psychology. All that came together to help companies. Initially, we did not know, how to make money with this model. But we were the first-of-its-kind analytics company in the world. There is a price to be paid to be ahead of the curve. But we had some very good clients which gave us interesting problems- like automated credit lending. We built the first statistical scorecard for ICICI for credit choices in India. It was revolutionary at that time. Then we helped Unilever to handle the 2008 recession- we used the science of data to carve the right offers for customers as per what data told us – on what would resonate well with customers. We cut our teeth into these areas with these customers. Soon analytics became more and more interesting. The idea, throughout, was — can we predict, and can these predictions help companies with decisions?

We are a formidable country. We are the only one today that is producing talent for the world.

What led to the move to AI?

The field of AI and analytics, themselves, have been evolving. When we started, neural networks were not fashionable. There were issues like the ‘black box’ challenge and explainability. It was true. For any dataset with less than a million rows, it is hard to be incremental with neural networks. In that age, these networks were also not deep. We went for classical techniques like random forests, decision trees, and regression. We used those algorithms to help companies make decisions. From the mid-2000s to late 2000s, AI started to pick up, especially in speech processing and computer vision. It started to move from structured (just rows and columns) to unstructured data like voice or text. We started making presentations to boards of companies. They also started getting interested and leveraging these approaches.

What kind of companies did you go after?

We had the 10-20-30 principle. Revenue 10 billion or 20 billion market cap or serving 30 billion in customers. We went after large Big Fortune 100-sized companies. This clarity on who to serve helped us build a really solid business between 2010 and 2016. By the pandemic, AI had taken off. Then we had the ChatGPT moment in 2022 and now everything has exploded to a new level. We have seen all stages of AI throughout our journey.

Have you changed in the era of Gen AI? In the last couple of years?

It is quite fundamental in the way it is shaping the world. It is just an addition to the AI wave. But this time, the intelligence is so much all-encompassing, stronger, and better. It is going to affect the world in 3 major ways- a rise in productivity, an explosion of intelligence in a more responsive and personalized way, and computing that is becoming human-like and user-friendly. We have changed in the way of understanding how big it is and how dramatically it will change the world- specially the problems we solve for customers. We have realized that one has to be really deep and invest in R&D to be ahead in this era. Like- foundational model work for text-to-image conversion or AI coach for coaching wisdom, and other many customer-facing AI products. We need sharper and sharper talent in this world which is changing so fast.

What are the industries which have changed the most with respect to AI?

We serve about 153 Fortune-500-sized global clients so we have a good view. Big changes are happening in places of cognitive and repetitive work. That’s where AI will make a huge dent. Like banking, creative tasks like image or video creation (now imagination can be translated into reality in a not-so expensive way and fast), legal firms, advertising firms and insurance players. It is augmenting humans – but with fewer hours and, hence, fewer people.

We had the 10-20-30 principle. Revenue 10 billion or 20 billion market cap or serving 30 billion in customers.

What about India- can we capitalise Gen AI to be a tech superpower?

We are a formidable country. If we take a view of the next ten years- We are the only one today that is producing talent for the world. In most G20 countries the net employment growth is negative. We are producing some really good engineers. Our role as a talent provider to the world- will make us successful. We are also the fastest growing economy of this size. If we continue to grow at 8-10% in the next ten years, we will be a very huge economy. What we need is an investment in talent, entrepreneurship, and technology. Talent, per se, could be net negative (with Gen AI) unless we can harness it and use it to imagine new stuff. We can train people to use these AI tools and be expert problem-solvers. We should also set up a better computing capacity for India- so that people with ideas but no access to GPUs can have this access. We also need to incentivize small companies to solve some government use-cases. If we do all that, our path is cut out to play a really important role ahead in the future.

Srikanth Velamakanni

Co-founder and Group CEO, Fractal Analytics

By Pratima H

pratimah@cybermedia.co.in