A new artificial intelligence model, Project Indus, launched on , may not immediately register on the radar of global tech investors. However, its focus – a multitude of Indian languages and dialects – signals a significant and often overlooked, trend in the development and deployment of AI: localization and inclusivity.
Developed by Tech Mahindra’s Makers Lab, Project Indus initially focuses on Hindi and its 37+ dialects. This isn’t about competing with the dominant, English-centric large language models (LLMs) currently shaping the AI landscape. It’s about extending the benefits of AI to a population largely excluded by existing technologies. According to Tech Mahindra, India is home to over 19,000 dialects, and most current LLMs are built for English, leaving millions without access to AI-powered solutions.
The project is underpinned by a collaboration between Tech Mahindra, Dell Technologies, and Intel. Dell provides the high-performance computing solutions, storage, and networking capabilities necessary for implementation. Intel contributes its infrastructure solutions, including Xeon processors, OneAPI software, and future generation products featuring Advanced Matrix Extensions (AMX). Tech Mahindra will also leverage Intel’s Gaudi AI Accelerators for future model training and to upskill its employees on Intel’s product portfolio.
The core of Project Indus lies in its ‘GenAI in a box’ framework, designed to simplify the deployment of advanced AI models for businesses. This is a crucial element. While developing a language model is a significant undertaking, making it accessible and easily integrated into existing enterprise systems is the key to widespread adoption. The model itself has 1.2 billion parameters, designed to be lightweight enough for smaller businesses and individuals to deploy without requiring expensive infrastructure.
The potential applications are diverse. Mahindra Finance is already exploring the use of Project Indus to provide financial services to low-income families in their native dialects. This highlights a key benefit: increased financial inclusion. The model also aims to enhance educational accessibility, allowing students to interact with AI-powered chatbots in their own languages. It offers small businesses and rural entrepreneurs the opportunity to integrate AI-driven solutions without language barriers.
Nikhil Malhotra, Global Head of Makers Lab at Tech Mahindra, described Project Indus as the company’s “seminal effort to develop an LLM from the ground up.” This suggests a long-term commitment to building indigenous AI capabilities tailored to the specific needs of the Indian market.
The launch of Project Indus coincides with a broader strategic move by Tech Mahindra. On , the company announced a partnership with NVIDIA to launch a new education-focused AI model, also under the Project Indus umbrella. This collaboration suggests a dual focus: expanding the model’s capabilities and targeting specific sectors, such as education, with tailored AI solutions.
The choice of Intel’s hardware and software is noteworthy. The use of Intel Xeon Scalable processors and AI PCs ensures efficient and localized performance. More importantly, the commitment to leverage Intel Gaudi AI Accelerators signals an intention to build a robust and scalable AI infrastructure for future development. This is a strategic decision, positioning Tech Mahindra to capitalize on the growing demand for GenAI expertise among its global customer base.
While the immediate financial impact of Project Indus remains to be seen, the broader implications for the AI industry are significant. It demonstrates a growing recognition that AI’s potential will only be fully realized when This proves accessible to a wider range of languages and cultures. The focus on Indian languages is particularly relevant given the size and diversity of the Indian population. This localized approach could serve as a model for other regions seeking to overcome language barriers and unlock the benefits of AI for their citizens.
The success of Project Indus will likely depend on several factors, including the accuracy and fluency of the model in various dialects, the ease of integration with existing systems, and the availability of sufficient training data. However, the project’s innovative ‘GenAI in a box’ framework and the strong partnerships with Dell and Intel suggest that Tech Mahindra is well-positioned to overcome these challenges. The initiative represents a significant step towards a more inclusive and equitable AI future, one where technology speaks the language of the people.
