On Friday, Krutrim Si Designs, the artificial intelligence (AI) venture led by Ola co-founder Bhavish Aggarwal, revealed its latest creation, Krutrim – a foundational large language model (LLM). This marks the company’s entry into the highly competitive AI landscape, where giants like Google, Microsoft, and OpenAI currently dominate.
Krutrim, the latest creation from Krutrim Si Designs, distinguishes itself by incorporating an expansive dataset that reflects the linguistic diversity of India. The generative AI applications embedded within Krutrim have been fine-tuned to cater to all major Indian languages, making it a noteworthy addition to the artificial intelligence landscape.
The model’s development was spearheaded by a team of accomplished computer scientists, strategically based in both Bengaluru and San Francisco. Their collaborative efforts have resulted in a robust and sophisticated large language model that showcases the fusion of expertise from two tech hubs.
What sets Krutrim apart is not just its technological prowess but its practical application. Beyond being a mere technological advancement, Krutrim is poised to play a pivotal role in Krutrim Si Designs’ ecosystem. Specifically, this model is set to power Krutrim’s conversational AI assistant. This AI assistant is designed to go beyond mere language understanding; it can fluently interact in multiple Indian languages. This feature marks a significant leap in enhancing user experience and accessibility, aligning with the vision of making advanced AI technology more inclusive and versatile.
“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.”
Named after the Sanskrit term meaning “artificial,” Krutrim emerges as a comprehensive family of Large Language Models (LLMs). Within this family, there are distinct entities such as Krutrim Base and Krutrim Pro, each designed with specific features and capabilities.
The Krutrim Base is the foundational model, setting the groundwork for the family. It serves as the backbone for the broader Krutrim ecosystem, demonstrating its capabilities in handling various linguistic nuances and complexities.
On the other hand, Krutrim Pro takes a step further with an array of advanced features. It boasts multimodal capabilities, indicating an ability to process and understand information from different modes, such as text and images. Additionally, Krutrim Pro is equipped with a larger knowledge base, showcasing an expanded understanding of diverse topics and domains. The technical advancements incorporated into Krutrim Pro enhance its inference capabilities, allowing it to draw more nuanced and contextually relevant conclusions.
The Krutrim family, with its various models, represents a significant stride in the evolution of artificial intelligence, combining linguistic understanding with multimodal features and advanced knowledge capabilities for more sophisticated inference.
Ola emphasizes that Krutrim’s exceptional linguistic capabilities position it as a valuable tool with a wide array of applications, spanning from educational uses to business communications. The model incorporates state-of-the-art techniques in safe AI, aiming to minimize inappropriate responses and ensuring responsible use in various contexts.
In addition to the development of Krutrim, Ola is actively engaged in building AI infrastructure. The company is working towards establishing an indigenous data center, with long-term plans encompassing server computing, edge computing, and supercomputers. This strategic move reflects Ola’s commitment not only to physical investments in data centers but also to advancing the technology behind them. The goal is to enhance the efficiency of data centers, reducing costs while simultaneously championing greener and more sustainable solutions.
Anticipating mid-2024 for prototype production and a full-scale rollout by the end of 2025, Ola’s timeline underscores its dedication to delivering cutting-edge AI technology to the market. Bhavish Aggarwal emphasized the importance of both physical infrastructure and technological advancements to ensure that data centers meet evolving needs, aligning with Ola’s broader vision for technological innovation and sustainability.
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.”
Ola’s venture into artificial intelligence with Krutrim represents a significant leap forward in the realm of large language models. Krutrim, the family of LLMs, showcases superior linguistic skills and versatile applications, making it a valuable asset across education and business communications. The incorporation of safe AI techniques underscores Ola’s commitment to responsible and ethical AI use.
Beyond Krutrim’s capabilities, Ola is strategically investing in AI infrastructure, including the development of an indigenous data center. This move extends to server computing, edge computing, and supercomputers, reflecting a holistic approach to technological advancement.
With a production timeline set for mid-2024 for prototypes and a full rollout by the end of 2025, Ola is poised to make a significant impact on the AI landscape. Bhavish Aggarwal’s emphasis on efficiency, cost reduction, and sustainability in data centers further highlights Ola’s commitment to technological innovation with a focus on environmental responsibility. As Ola pioneers advancements in AI and infrastructure, it sets the stage for a transformative era in the intersection of technology and sustainability.
