Ola’s Chief Launches India’s First AI Language Model Krutrim: Here’s What We Know

Ola runs cabs and other services in India but the company has ambitions far beyond the aggregator space. Ola’s chief Bhavish Aggrawal has jumped into the AI ring with one of India’s first indigenous AI models called Krutrim. Ola claims Krutrim is built on local Indian languages, data and more.

The company wants the country to ditch the Western AI products (ChatGPT and Google pointed here), as India will be building its own AI model. Ola had announced Krutrim (which means artificial in Sanskrit) during a live event on Friday, where Aggrawal was proudly talking about its local capabilities and what it could bring to the table in the near future.

The interface of Krutrim looks fairly simple and user friendly, which could be one way to attract more people in the country to use the AI chatbot. He also mentioned that the model can understand 20 Indian languages and can generate in 10 Indian languages, including Marathi, Hindi, Telugu, Kannada, and Odia, among others.

During the event, Ola’s chief showed a quick demo of the AI chatbot, putting it across simple queries like writing stories, poems and even welcoming the guests who were present at the launch venue. Krutrim will have two models, the basic one and the Pro model, the latter will be coming out in the next quarter, Ola announced.

Krutrim has been built by training over 2 trillion tokens, and the company admitted that the AI model is still in its beta phase, which is why the tests were limited to basic responses for now. The company also wants to build a full AI stack that includes an AI chip, sustainable AI cloud, family of AI models and the user-facing application.

AI is undoubtedly the toast of the town these days, and the year 2023 clearly shows us the path in which AI is headed in the coming years. Most tech giants like Google, OpenAI, Microsoft and Meta among others have entered the AI space with their respective LLMs but Ola is a surprising addition to this roster, looking to cement India’s position and direction on AI along with the government.

However, running complex AI models requires heavy investment, not only in terms of the computing power, but also the funds needed to sustain the applications that work through their machines.

OpenAI used Microsoft’s big pockets to develop ChatGPT in the past 12 months, while Google continues to build its AI roadmap with the Gemini series. Ola will need similar impetus from the investors to keep the computers running, and that’s where this demo of Krutrim might help its case.

But only time will tell if Bhavish and Co. have a model that can succeed in the long run. After all, AI cannot be just about a business model that thrives till the point investment flows in, it needs a long term approach that will help it evolve and become more than just a language model for consumers as well as businesses.