UPI Moment For Compute, An Indian Fintech LLM, Sarvam's First Demo: Key Takeaways From Adbhut India

A digital public infrastructure approach to computing, an India-focused financial services large language model, and Sarvam.AI's first-ever authoring experience demo were among the key takeaways as India's leading technology thinkers, led by Nandan Nilekani, demonstrated a way forward for the implementation of artificial intelligence at scale in India.

At Adbhut India, an event backed by Nilekani's EkStep Foundation, several innovative use-cases and infrastructural progress to support India's population and businesses in adopting AI were launched.

The event marked the launch of Open Cloud Compute, a project to enable access to compute infrastructure.

Open Cloud Compute is launched by People+ai, an initiative of the not-for-profit EkStep Foundation. It announced a partner consortium of 24 organizations that would come together to facilitate large-scale computing access.

Experts have dubbed the project a UPI moment for cloud compute access. Going beyond reliability at a few large data centres currently in the country, the open cloud commute project will expand the network of providers for companies of all sizes and provide cheaper costs.

"Over the next few years, demand for compute will go up due to increased digitalisation and AI, and the global cloud compute market is expected to triple in size," said Tanvi Lall, director of strategy at People+ai.

"Through OCC, we are taking a digital public infrastructure approach to unlocking access to compute at scale. The interconnected micro-cloud computing infrastructure aims to unite numerous independent providers on a single network, enhancing their discoverability and utilisation. Users of this infrastructure would have access to computing power and related services from many providers," she said.

"It will enable many microcomputing providers, potentially in the hundreds, to unite and create a megacomputing network powered by trust. As Nandan once said, micro is the new mega," said Dr. Pramod Varma, chief technology officer at EkStep Foundation.

At the event, Setu, in collaboration with Sarvam.AI, launched Sesame— India's first domain-specific large language model aimed at providing better financial services.

Sesame, being seen as a ChatGPT for financial services, showcased a demo of how e-commerce and financial products can be hyper-personalised based on data accrued from account aggregators. It was tested on WhatsApp as a chatbot by Nikhil Kumar of Setu, a fintech that was acquired by payments giant Pine Labs.

Sarvam.AI's Pratyush Kumar demonstrated how the indigenous AI research firm's authoring experience functions. Kumar gave a low-down on how chatbots are created and why the cost of AI compute needs to be around Re 1 per transaction.

To bypass the state and quality of literacy, the chatbots it is developing are focused on voice-led interfaces that seamlessly understand dialects and "get things done", Kumar said.

Sarvam.AI is one of India's top startups in the AI space and one of the best-funded too. It raised $41 million, or Rs 340 crore, in December last year and was founded by Vivek Raghavan and Kumar.

Raghavan has been working for a decade on building digital public infrastructure at the Unique Identification Authority of India, which oversees Aadhaar. While Kumar worked at AI4Bharat, a Nandan Nilekani-backed initiative of IIT Madras, which is building open-source language AI for Indian languages,.