As governments across the Global South move from pilot projects to real-world deployment of artificial intelligence, experience is becoming more valuable than ambition. For Nakul Jain, Managing Director and CEO of Wadhwani AI Global, India’s long engagement with AI in public systems offers practical lessons for countries such as Rwanda. Speaking to The New Times, in an exclusive interview on the sidelines of the AI for Impact Global South Forum 2026 in Kigali, Jain drew on nearly a decade of work embedding AI within government systems in India and explained why many of the same lessons apply across Africa. How Wadhwani AI began Wadhwani AI was founded in 2018 by two Indian brothers who had grown up and studied in India before moving to the United States. At the time, Jain said, artificial intelligence was largely driven by big technology companies, while governments, particularly in developing countries, lacked access to the infrastructure, skills, and institutional capacity needed to use AI effectively. The organisation was created to address that gap by focusing exclusively on public-sector applications of AI. India became its first area of operation. Rather than working as an external vendor, Wadhwani AI embedded itself within government systems, working closely with ministries to identify problems, design solutions, deploy tools, build internal capacity, and evaluate outcomes over time. India’s experience at scale Over the past seven to eight years, Wadhwani AI has worked with India’s ministries of health, education, agriculture, urban affairs, women and child development, electronics, and information technology. That collaboration has resulted in between 20 and 25 AI solutions deployed across multiple states. Some of those systems now serve a combined user base of about 150 million people, Jain said, but scale alone does not tell the full story. He noted that many AI applications perform well in controlled settings but struggle once deployed in real-world environments. Why context determines adoption According to Jain, poor adoption is often the result of insufficient contextualisation. Language barriers, cultural differences, unsuitable data, and unanticipated changes in workload can all undermine otherwise well-designed systems. “When AI systems make incorrect decisions, it is the end user who bears the consequences,” he said, adding that this is why responsibility must be built into every stage of development. Those lessons have shaped Wadhwani AI’s emphasis on responsible AI, including clear data governance, transparency, and ongoing monitoring after deployment. Expanding beyond India In 2025, Wadhwani AI launched its global programme, extending its work beyond India to other parts of the Global South, beginning with Africa. The organisation engaged with governments in Kenya, Rwanda, Ethiopia, Nigeria, and Zanzibar to better understand local priorities and institutional readiness. Rwanda, Jain said, stood out during that process. He described the pace of government decision-making and the level of preparation around digital governance as notable, pointing to the existence of a national AI strategy and ongoing programmes already aligned with public-sector priorities. South–South collaboration and data sovereignty Jain cautioned that much of today’s AI infrastructure and innovation originates in the Global North, meaning many solutions are designed for Western contexts and may not translate well elsewhere. He said this also raises concerns around data sovereignty and long-term dependence. For him, South–South collaboration offers a way forward. Based on Wadhwani AI’s experience, he estimates that 70 to 80 per cent of the public-sector challenges addressed in India are also relevant in countries such as Rwanda. That overlap creates opportunities for policy exchanges, joint training programmes for decision-makers and practitioners, and shared development of AI systems that can be adapted locally rather than rebuilt from scratch. Use cases across key sectors Jain pointed to examples from India where AI is already being used within public systems. In healthcare, AI tools support tuberculosis screening, maternal health risk prediction, and early identification of potential disease outbreaks. In education, AI is used to assess foundational literacy and numeracy, identify students at risk of dropping out, and assist teachers in targeting interventions. In agriculture, farmers use AI-powered tools to identify pests and crop diseases and access advisory services in local languages. He said these examples demonstrate the breadth of potential applications when AI is aligned with public needs. Infrastructure and long-term impact Compute infrastructure, Jain added, is another area where collaboration could be beneficial. While most global AI compute capacity remains concentrated in the Global North, India has invested in building domestic capability, and Rwanda has begun laying groundwork in this area. “When countries work together, they can reduce costs and strengthen their negotiating position,” he said, noting that delays caused by overly complex processes can slow progress in a rapidly evolving field. For Wadhwani AI Global, success is measured by sustained use and reach rather than pilots. Jain said the organisation aims to impact half a billion people across India and the wider Global South by 2031, by embedding AI into public systems and strengthening institutional capacity.
