Harnessing AI’s Potential in India with Srikanth Velamakanni
How AI is reshaping industries in India—from healthcare to customer service and beyond
From transforming healthcare to reshaping customer service, AI is making a tangible impact across industries. In this episode, Srikanth Velamakanni, Co-Founder and Group Chief Executive of Fractal Analytics, discusses how AI is augmenting human capabilities and driving real-world outcomes. What does this mean for businesses in India? How can India leverage AI responsibly to stay ahead globally? Srikanth and Govindraj Ethiraj tackle these questions and dive into India’s role in the evolving AI landscape, as well as the ethical and societal implications of this powerful technology. Tune in for an in-depth look at AI’s future in India.
TRANSCRIPT
NOTE: This transcript has been done by a machine. Human eyes have gone through it but there might still be errors in some of the text, so please refer to the audio in case you need to clarify any part. If you want to get in touch, do drop us a message on [email protected].
---
Govindraj Ethiraj: Welcome to another insightful episode of Nasscom Conversations!
Nvidia CEO Jensen Huang’s latest visit to India has fired up our collective imagination about AI. If Huang got a rockstar reception, its because AI today offers similar and exciting potential.
Lets recap a little. In the last few years, AI has moved from being a behind-the-scenes enabler to something much more mainstream—thanks of course to developments like ChatGPT and generative AI.
But with all the excitement around AI, where does the reality stand?
And what impact is this having on businesses, particularly the consumer facing ones ?
And and most importantly, will AI put us out of our jobs ? Not quite, if you are…well, stay tuned as I speak with Srikanth Velamakanni, Co-Founder of Fractal Analytics.
I kicked off by asking Srikanth to take us through how AI is augmenting human intelligence, automating processes, and, in some cases, even exceeding human capacity.
From healthcare to customer service, the applications of AI are endless. We'll also explore how enterprises in India are embracing AI to transform their businesses and stay ahead of the curve in this evolving landscape.
Srikanth Velamakanni: AI has been around for 70 plus years, almost 70 years, 1956, it was coined during a summer project, Dartmouth Summer Project. For the first 50 years, not a lot of action, even though some of the work has earned Nobel Prizes, as of last week, you saw two Nobel Prizes being awarded to AI scientists, Geoffrey Hinton for his work around creating deep learning networks, which he did in the 70s and 80s, and then the work done by Demis Hassabis and his team at DeepMind during the 2014 to 2020 era.
And then it took off between 2010 and 2020 in a way that, yes, the common man probably didn't see it, except that they were being experiencing it. For example, if you watch a YouTube video and it's very personalized, and there's the homepage on the first three, four tiles, you are all clickable. That's because it's taking a billion videos and picking out the three that are relevant to you right now at this point in time.
Govindraj Ethiraj: We’ve seen AI rapidly evolve, particularly with its impact becoming more visible in everyday applications. A recent example is the use of AI in the healthcare sector, where researchers have developed AI-driven models to predict pandemics or assist in drug discovery. One such breakthrough is the use of AI by DeepMind’s AlphaFold to solve protein structures, which could revolutionize drug discovery.
Srikanth speaks of instances where AI’s impact is becoming tangible, and how these advancements are changing industries in real time?
Srikanth Velamakanni: It is one of the favorite topics for everyone because we know healthcare is broken around the world and it needs to be fixed. So just about a week after ChatGPT was launched, there was a guy called Pete Cooper.
His Twitter handle was at Pete Cooper. He basically talked about how his dog was saved with AI. This is very interesting because his dog was anemic.
It was about to die. He had taken the dog to a number of vets and the vets had concluded that nothing can be done. We can't figure out what's going on.
It's anemic, it's losing blood. And then he was about to give up. Then he said, there's something called ChatGPT which has been launched and GPT-4 was launched that week.
So he took all the medical reports and put it into GPT-4 and said, tell me what could be happening to my dog. And while GPT-4 clarified that, look, I'm not a vet. I don't necessarily know all of this stuff.
It gave him two important hypothesis. He took this report, went back to the vet and said, this is what my GPT-4 is saying. Does it sound sensible?
And the vet says, yeah, this is one of the things I did not check on but maybe GPT-4 is right. And literally within a couple of weeks, the dog is all right. So this is an example that the human beings, we have some strengths but we have a number of weaknesses like we have low attention span.
We cannot process too much information at the same time. We get distracted. We have other priorities.
It's really, really large. I may be getting the number wrong by a bit but it's safe to say that no doctor in the world can read all the papers published in their specialty in a single year. It's impossible for them to be practicing and doing this at the same time.
Govindraj Ethiraj: Healthcare is a major area where AI is already making a real impact, like helping doctors analyze X-rays by comparing them with thousands of other cases.
Beyond healthcare, AI is also transforming backend operations in industries, from service configuration to manufacturing. How is driving change in these areas and what are the bst examples here?
Srikanth Velamakanni: Take an example of any customer service center. Now, if you, let's say your car broke down and you wanted to call and figure out what are the terms and conditions of your insurance plan and what is eligible, et cetera. Now, when you make this phone call, it takes a long time for the customer service agent to understand your plan limitations and give you precisely the right answers very, very quickly.
So what now is happening is we are creating systems which are called agent assist.
So the agent who's still answering my questions but the agent now has a search interface where she or he can quickly get the right information and then answer the questions, reduce escalations, reduce the call handling time and reduce the error rate dramatically from 10% to almost 2%.
One of the fastest-growing use cases now is making sure all the agents, contact centers that we all talk to for various kinds of problem-solving, they're all using AI in order to make their lives better. So one example is Fractal does this thing for HDFC Bank. It's called EVA, the chatbot that they use.
It does the work of 1,500 agents already. The number of questions it answers, it augments or it substitutes for 1,500 workers, and it generates 50,000 leads for HDFC Bank every month. That's a very powerful system.
Govindraj Ethiraj: We’ve discussed how AI is transforming industries, but what of augmented intelligence? How does it complement AI, and what are some real-world examples where this convergence is enhancing human capabilities?
Srikanth Velamakanni: See, there's practically no difference, except that AI, when it's used in order to improve human performance, I would call it augmented intelligence. So AI, when designed to be autonomous, doing its own thing, because it's doing a part of a job or it is automating a part of a task is automation. But when it is used specifically designed to interact with human beings in order to drive human performance to the next level, I think that's augmented intelligence.
Elon Musk is driving one version of this, which he's calling as Tesla's autopilot system. Well, the idea is that it does most of the job, but you have to pay attention. It's learning, it's driving most of the time, but you need to be in attention.
You still have a steering wheel, you still have to take over when something goes wrong and drive the car to safety. That's the expectation of Elon Musk's technology.
It's because it's not level four, level five certified. Now, autonomous driver driving comes from level one to level five. Level one is a basic automation.
Level two is already sort of doing most of the job, and level three, four, and five. Now, level three is where Tesla is, where it does the driving most of the time, but the human being is still the driver. He or she has to make the decisions in a crunch situation.
Now, level four is the first time that it says you don't need the human being to drive the car at all. In fact, what Waymo is doing, it's Google's subsidiary, Waymo is doing, is it's saying that you don't need a steering wheel.
It's already doing 100,000 rides a week in three cities, in San Francisco, in Los Angeles, and Phoenix. So it's thinking is that, let me build one very good driver, the software driver. They're thinking of building one driver, and the driver drives the car.
Now, Tesla's vision is still augmented. They're saying it'll do the job 95% of the time. The trick with that is when you design a system for working 95% of the time, and you expect the human beings to smoothly take over 5% of the time, it has to be designed better than where the Tesla is right now.
Govindraj Ethiraj: AI innovation is progressing rapidly, whether it's in Silicon Valley or India, but the success of AI also depends on how effectively humans can use it to drive business outcomes. Given that the human aspect is often local and specific to industries, how well-prepared are we on that front? What do we need to do to improve our ability to harness AI effectively and ensure better business outcomes?
Srikanth Velamakanni: I think this wave of AI is saying, look, I will meet you where you are. I'll understand whether you're speaking in English or Marathi, whether you're speaking in full sentences or half sentences, whether your thoughts are clear or not clear, whether your questions or doubts are clear or not clear. I'll make the best use of this and then help you out from where you are to where you want to go.
Therefore, it will be helpful for people to understand where the AI tools are, play around with them and get better. Now, as NASSCOM, what we have to do is that we have to make everyone in our country AI literate. It'll be very useful for us to make everyone AI literate, not in the way of understanding the mathematics behind it.
And the biggest advances that are happening are in agentic systems and reasoning. If I can have a million people understand how to build AI agents that can automate and augment tasks, we know that India will leapfrog to the next level of this IT wave.
Govindraj Ethiraj: As India positions itself as a key player in driving global AI research, what specific areas are emerging as major opportunities for innovation? Where can India take the lead, and what doors are opening for us in the AI landscape?
Srikanth Velamakanni: The biggest, so one is this whole area of building foundation models. That's, these are the basic building blocks of all AI. And the race is between OpenAI, Google, Anthropic, and a bunch of other companies.
They need access to hundreds of thousands of GPUs. They need extraordinary talent. And then they need patient capital to make all of this work.
So that is at a scale which may not be easy for India to aspire to. But there is a next step, which is around using these foundation models, which could become like utilities. Eventually they'll become so commonplace, easily accessible that you don't even have to worry about how they're getting produced.
So let's say we are all trying to build a website. Now, can you have a developer, a tester, a debugger, somebody who is project manager, a product manager, multiple roles of people work in creating, let's say a website. But what if I can replicate that with AI agents?
That's something that'd be very powerful for let's say the Indian software industry, if we can use AI agents to build the new wave of software that we sort of built for the whole world. So that's one area. Second area is to bring more reasoning capabilities into the AI systems.
Govindraj Ethiraj: As India continues to take a larger role in the global AI landscape, both within the country and internationally, what are the key ethical and societal considerations we need to address? How can we proactively prepare for these challenges and ensure that India stays ahead in creating responsible AI systems?
Srikanth Velamakanni: Great power comes great responsibility. And the idea of that is AI fundamentally has to be safe.
So if AI, we give a lot of, let's say, autonomous capacity to AI, what if it can do something really harmful? And how do you make sure that you can stop it or you design safety into the way it is built in the first place? So this is a whole area called AI safety, which is to make sure that AI never goes beyond its limit and doesn't do any damage.
These are all the kinds of things you have to look at, which are part of a system called Responsible AI. Now, this is an area that is very, very crucial. NASSCOM has done a lot of work in building the Responsible AI recommendations for India, but India needs to incorporate a full-fledged Responsible AI principles, guidelines, and toolkits.
And then another way of doing it is to make sure that some of the most clear, the very powerful AI is not let out without clear regulatory approvals and so on. So for the basic AI, self-certification, but for extremely powerful AI, there may be a time that might come when we may have to make sure that this kind of AI doesn't escape the lab, and so to speak.
Govindraj Ethiraj: That's a very, I mean, it's a very vivid example of something escaping the lab and going loose. Okay, so coming back to India and the efforts that we're putting in today, are there, you know, if you were to think of, let's say, a few initiatives or partnerships or cutting-edge initiatives or partnerships going forward as an industry body, as a country, what would they be at this point of time?
Srikanth Velamakanni: Think about AI as like nuclear technology back in the 1940s and 30s.
So we need, at some point of time, global cooperation on very powerful AI. If that kind of AI is not yet there, I think today's AI is relatively primitive compared to what might happen in the next, let's say, three to four years, maybe five years. Even before we get to that place, is it possible that various countries in the world can come together and say that we can create this international atomic energy agency, like platform, so that some basics are guaranteed?
For example, that we will not use AI in autonomous weapons. Unfortunately, the train may have left the station on that one. It's already, my guess is that many countries in the world are already building AI into their autonomous weapons.
Can we restrict AI from being part of, let's say, bioweapons or hacking or cybersecurity or cyber attacks? Can we make sure that at least some places we are controlling the kind of dangerous AI that could be out there?
Aviation is one of the safest industries. Just imagine, aviation could have been very risky, but we made it very safe because of very proactive safety considerations in the way we designed these planes and we designed the systems around these planes, including air traffic control.
What if we can design that for AI as well? That's a huge, huge possibility, which can happen with global coordination between various kinds of countries and various kinds of nonprofit bodies.
Govindraj Ethiraj: So, you have the International Civil Aviation Organization which does all of what you said, or a lot of what you said. Do you see something similar evolving in AI and does India have a role to play in designing that?
Srikanth Velamakanni: India has been asking for it. There’s the GPAI, which is a global partnership on AI. India led it last year. These kinds of questions have been suggested, but it’s still not broadly agreed. It seems like an arms race, where countries fear that slowing down might allow others to get ahead.
We haven’t had a “Hiroshima moment” that unites global action. But we must continue the dialogue to see if India can play a role. We’ll be more credible if we build powerful AI ourselves. Historically, only the powerful get heard.
One of my recommendations is to build powerful AI in India. Just as we did with ISRO in the 1950s, we should create a department of AI to lead research into AGI. As we approach AGI, the risks become bigger, and we need to be prepared.
–
Srikanth Velamakanni: As CEO of Fractal, I'm very involved in our AI research. Recently, we’ve presented two papers at NeurIPS, the premier AI conference. It’s exciting because not just Fractal, but many Indian companies are doing cutting-edge AI research, and we have a huge opportunity with the talent we have in the country.
We also launched Collider on AI, a free text-to-image product, and Vaidya.ai, which I’m particularly excited about. Vaidya is like ChatGPT for healthcare, and it could become part of a national platform to address India’s healthcare challenges. AI solutions like these can really help improve outcomes, especially given the scale of the problem we’re tackling.
Govindraj Ethiraj: So, if I can supplement that with another question. So, one of the things that your own companies do or one of your companies does is scan X-rays for tuberculosis symptoms. Now, that's an example of, let's say, how healthcare problems can be resolved or addressed at scale, particularly remotely in countries like India.
Is there any other such problem that you're thinking of or wanting to bring technology to bear?
Srikanth Velamakanni: So, Cure.ai did that for tuberculosis. It's game-changing.
I think it's still not being implemented across the country. It has a real opportunity to reduce fatalities in India through tuberculosis. It is one of the largest killers in the developing world.
We have reached 20 million people around the world, but 20 million is a drop in the bucket compared to the 1.4 billion people of our country and 8 billion people on the planet. So, I would ideally like it to reach everyone. Secondly, we looked at lung cancer and looked at how we can diagnose lung cancer.
The third thing, this is very new. And since the Nobel Prize of the last couple of weeks and the work that DeepMind has done, one of the ideas that I have for India is what if we have an AI-native healthcare company that is doing drug discovery for challenges that India needs to solve?
Govindraj Ethiraj: In the broader debate between augmented intelligence and automation, how do you think young engineers and those entering the job market can best align themselves with the opportunities in this space? How can they prepare themselves without feeling overwhelmed by the rapid advancements?
Srikanth Velamakanni: Yeah, first of all, I would say that for young students, it's very important to get the basics of your engineering and math education. If you are a budding engineer in the AI space, it's still important to learn probability, statistics, linear algebra, and basics of coding. Even though a lot of tools have come in, you still have to learn them because you have to be even better than those tools and use those tools effectively.
So that's number one. Number two, I would absolutely invest in learning the basic code assistant tools that can help you write code using AI. These have become very popular and they can improve the coding productivity by 50 to 100%.
Govindraj Ethiraj: That brings us to the end of today’s conversation with Srikanth Velamakanni. We’ve explored the vast impact of AI, from transforming industries like healthcare and customer service to how it augments human intelligence and drives innovation in real-world applications. Srikanth shared insightful examples of AI saving lives, streamlining business operations, and even enhancing decision-making in finance and customer support.
As AI continues to evolve, the balance between augmented intelligence and automation will shape how we harness its full potential.
What’s clear is that India has a significant role to play, not only in advancing AI technology but also in ensuring its responsible and ethical use. For young professionals entering this space, now is the time to build foundational skills while staying curious and adaptable in a rapidly shifting landscape.
Thank you to Srikanth for his valuable insights, and thank you to our listeners for joining us. Stay tuned for more conversations on the cutting edge of technology with Nasscom Conversations.
