The roadmap to extraordinary opportunities in the AI age

AI is making heads turn. Avataar Venture Partners’ Annual General Meeting made for an insightful discussion on India’s place in the age of AI and the opportunities that lie ahead.  

The world has come far in the quest to make machines as intelligent as humans. However, Artificial Intelligence (AI) as a technology is still largely a ‘research problem’, according to a panel of tech investors and entrepreneurs at the Annual General Meeting gathering of Avataar Venture Partners in Bengaluru last month. 

India is still at a nascent stage as far as research and infrastructure-building are concerned. 

‍

“The day AI is understood and it becomes an engineering problem – where people know precisely what has to be done, how it has to be done and built, and there's stability in terms of the underlying infrastructure – then I'd say the advantage shifts to India,” Shekhar Kirani, Partner, Accel, said during a panel discussion on impact and opportunities presented by AI in India. 

Globally, the tech industry has taken over academia in producing new models and building state-of-the-art AI, according to Stanford University’s 2023 AI Index Report. In 2022, there were 32 significant industry-produced machine learning models compared to just three produced by academia.

‍

Hemant Mohapatra, Partner at Lightspeed India, believes VCs can play a role in empowering India to get ahead in AI research by investing in companies that are building cutting-edge tech in India to attract talent back to India – both in startups as well as more mature companies.  

‍

India’s right to thrive 

The biggest and surest of opportunities emerge when changes in technology, access, and business models converge. “So this (the AI wave), for me, is a massive tsunami where all three are changing at the same time,” says Shekhar. 

‍

He is bullish that India will leapfrog the rest of the world when three factors align: 

• The access and ability to create proprietary data 
• When AI becomes an engineering-oriented problem
• Personalization to domain-specific vertical

‍

Besides data, Shekhar asserts the biggest moat lies in personalization. “If the AI can understand 'you' deeply, your travel habits, diet preference, likes and dislikes, and when it plans for you and serves you far better than anything else – because the data has been working back and forth, and you're giving feedback – that is the real moat because that cannot be transported to a newer company easily. Whoever gets access to that personalization layer, will win,” he adds.  

‍

For investors and entrepreneurs alike, AI presents an extraordinary opportunity to disrupt the existing pattern as well as to invest and create extraordinary companies. 

However, Hemant highlights that the deck is stacked against newcomers in every single wave and that incumbents usually have an advantage. 

“If you don't have an edge, incumbents have an advantage. The biggest companies before the internet were Microsoft, Oracle, and SAP. After the internet, they are still the same. So the challenge is on startups to build something so special and so powerful that it can disrupt the market,” he says. 

‍

Gen AI for the next billion people

When it comes to the application of AI, especially for enterprises at scale, one could be optimizing the operations, helping customer service agents, or automating part of their work. However, it must be deployed safely, ensuring the guardrails, monitoring, and controls are in place as well as continuous testing of AI deployed across the enterprise, enabling them to take action when they need to keep biases in check. 

However, Vivek Raghavan, Co-founder, Sarvam AI, notes that Gen AI today is heavily focused on the English-speaking world even though the tech itself is a ‘deflationary force which can change things in society’. At Sarvam, he is focused on enabling access to Gen AI to the next billion people by allowing them to speak in their native language to get things done. That will naturally expand the Total Addressable Market (TAM).

The key, he says, is to figure out ways to craft existing capabilities to create a new experience so that the cost of starting from scratch does not become a hurdle. 

“We believe India is an extremely cost-sensitive market. Hence, the price has to be very different which can be achieved by focusing on building smaller models, and focusing on cutting-edge technology, among others, to make Gen AI accessible at a very different price point. Once done, this may become something exportable to the rest of the world as well,” he explains. Hemant adds that cost can eventually be recouped through large inferences from users. 

‍

It is also important to maintain and retrain the models along the way because the world adds as much data as has existed in the last 20 years of the internet every year. 

Is AI in a hype cycle? “Yes but necessary,” replies Hemant, before adding, “Hype cycles are needed to push markets forward. But more than anything, investors believe it is an exciting time for VCs to join a time of technical evolution where one could enter, play that wave, and have it also be around when retiring from the workforce.