'India's own AI': Ola's Bhavish Aggarwal joins the race with Krutrim

Krutrim Si Designs, the artificial intelligence (AI) venture of Ola co-founder Bhavish Aggarwal, on Friday unveiled Krutrim, its base large language model (LLM).

Krutrim has joined the increasingly-competitive AI race dominated by players such as Google, Microsoft and OpenAI.

It has been built with the largest representation of Indian data used for its generative AI applications for all Indian languages. It has been trained by a team of leading computer scientists, based in Bengaluru and San Francisco.

This model will also power Krutrim’s conversational AI assistant, which understands and speaks multiple Indian languages fluently.

“AI will define the future paradigms of economy and culture. To become a true leader of the world, India needs to become a global leader in AI. Today, all AI models called LLMs are trained largely in English. But language is not just text but also the vehicle for cultural values, context and ethos. Due to India’s multicultural and multilingual context, the AI models just can't capture that. It needs to be trained on unique data sets specific to us. It also needs to be accessible to India, with India-first cost structures. An AI-first economy will improve labour and capital productivity. It will push the Indian technology industry on a nonlinear path and make it a global knowledge centre, a leader in scientific discoveries and a tool for cultural expression,” said Aggarwal, at the company event. 

He added, “With that vision, we have introduced Krutrim, the country’s own AI for 1.4 billion Indians. We are extremely excited to launch India’s first complete AI computing stack, (Krutrim), which understands our unique cultural context, connecting our future to our roots. With an India-first cost structure, Krutrim will have the largest representation of Indian data, enabling us to create novel models beyond LLMs across sectors. It will make India the most productive, efficient and empowered economy in the world.”

Krutrim, meaning “artificial” in Sanskrit, is a family of LLMs, including Krutrim base and Krutrim Pro, which will have multimodal, larger knowledge capabilities, and many other technical advancements for inference.

Trained on over 2 trillion tokens, Krutrim accomplishes better performance on multiple well-known, global LLM evaluation benchmarks, including MMLU, HellaSwag, BBH, PIQA and ARC.

During the launch event, Aggarwal demonstrated an AI chatbot that is powered by Krutrim. It functions similar to OpenAI’s ChatGPT and Google’s Bard.

Ola’s Krutrim model can switch between languages and discuss nuanced topics ranging from poetry in Bengali, to Bollywood movies, and creative masala dosa recipes.

It will be available in beta version from January 2024 as an application programming interface (API) for enterprises and developers, seeking to create AI-driven assistants that can converse in multiple Indian languages. Krutrim Pro will be available in Q4 of FY24. 

Ola said Krutrim’s superior linguistic skills make it a valuable tool for a wide range of purposes from education to business communications. It incorporates the latest techniques in safe AI to reduce inappropriate responses. 

The company is also working on AI infrastructure to develop an indigenous data centre and eventually, server computing, edge computing and super computers.

Production is scheduled for mid-2024 for prototypes and a roll out by the end of 2025.

“We're also building our own technology for data centres. It is not just the physical investment into the data centre, but also the technology. It’s very important to make data centres more efficient to bring down costs and also to make them greener and sustainable solutions,” said Aggarwal.

He further said, “While AI is the soul, the infrastructure and silicon is the body in which it runs. In India, we need to design our own silicon chips for building this.”