Krutrim AI start-up making weekly improvements despite criticism
Krutrim, an AI start-up founded by Bhavish Aggarwal of Ola, is reportedly making weekly improvements despite facing user criticism since its public beta launch in February. The criticism revolves around translation errors, factual inaccuracies, and logical reasoning issues of its models. Ravi Jain, Head of Strategy at Krutrim, shared this positive update during the Business Standard Manthan event in New Delhi.
Addressing AI challenges through user feedback
Jain acknowledged that hallucination problems are a common occurrence with any AI model. He emphasized the importance of user feedback from the beta program in refining the model over time. To tackle inaccuracies, Krutrim is focusing on enhancing its models with accurate and relevant data sets, utilizing all pertinent and factual Indian open web data for training its models.
Journey to becoming India's first AI unicorn
Krutrim made headlines as India's first AI unicorn in January this year, after securing $50 million in its first funding round from investors including Matrix Partners India. The start-up was initiated in April 2023 by Aggarwal and Krishnamurthy Venugopala Tenneti, a board member of ANI Technologies which owns Ola and Ola Electric. This achievement marked a significant milestone for the Indian tech industry.
Public beta reveals intriguing usage trends
The public beta of Krutrim has revealed interesting usage trends across sectors such as education and content creation. In Tier-II and Tier-III regions, multilingual applications have gained significant popularity. Examples include government employees using it to write letters in English or small business owners creating their catalogs, as shared by Jain.
Krutrim strives for advanced models and infrastructure
Currently supporting over ten Indian languages including English, Hindi, Tamil, Bengali, Marathi, Kannada, Gujarati and Hinglish, Krutrim is now striving to develop more advanced models for its platform and establish compute infrastructure. Jain emphasized the need for speed and resources to reach a satisfactory level of development in this AI-driven venture.
