AI ancient manuscripts are about to be read at a scale India has never attempted. The Ministry of Culture will use artificial intelligence and machine learning to decipher, digitise and preserve more than one crore ancient manuscripts under its Gyan Bharatam mission, The New Indian Express reported on 7 November. Business Standard carried the story on 8 November. The plan centres on an AI-powered National Digital Repository with a Talk-to-Manuscript feature for questions by voice or typing. The manuscripts hold centuries of writing on science, medicine, philosophy, literature and law, and few people today can decipher their scripts.
Gyan Bharatam, the AI ancient manuscripts mission, was announced in the Union Budget 2025, and Rs 491.66 crore has been sanctioned for the mission for 2025 to 2031. More than 1.25 crore manuscripts have already been reported on the portal launched in September 2025, and 11.53 lakh of them have been published so far. The portal already lists early AI tools: Gyan Mitram, an AI knowledge interface; Pandulipi Darpan, for script recognition, transliteration and translation; and a BharatiyaGPT conversational layer. A technology firm is being engaged to build the AI platform behind the repository.
AI ancient manuscripts and the Gyan Bharatam plan
The AI ancient manuscripts work will start with the page itself. Advanced optical character recognition and handwritten text recognition will turn manuscript images into searchable digital text while keeping the original structure, including headings, marginal notes and illustrations. The AI layer is expected to transcribe handwritten pages, translate texts and help identify their authors. It will also trace the provenance of manuscripts, reconstruct faded passages and find links between manuscripts held by different institutions and private collectors. The ministry has set one design rule for the whole system: grounding. Every AI-generated response on the platform will have to be backed by validated manuscript records and include page-level citations and quotations from the original sources.
Prime Minister Narendra Modi underlined the scale of the task at the International Conference on Gyan Bharatam at Vigyan Bhawan in New Delhi. He said India holds one of the world’s largest collections of manuscripts, with nearly one crore ancient works documenting different aspects of human civilisation and the evolution of knowledge. The conference, organised by the Ministry of Culture, brought together scholars, historians, conservationists, technologists and policy experts, and it saw the launch of the Gyan Bharatam Portal. Modi also said digitisation could help prevent intellectual piracy and unauthorised use of traditional knowledge, and set out four pillars: preservation, innovation, addition and adaptation.
AI ancient manuscripts in the laboratory
Indian institutes are already building the tools the AI ancient manuscripts mission needs. Researchers at IIT Jodhpur have developed an AI-based platform that restores damaged text, recognises old scripts and makes old knowledge accessible to scholars. The work, led by assistant professor Dinesh Mohan Joshi, combines AI, computational linguistics and Indian knowledge systems. The system first examines a manuscript to identify its likely script, language and condition, then enhances the page image, separates lines and words, and uses handwritten text recognition to identify characters. It has been designed to recognise six scripts: Devanagari, Sharada, Grantha, Nandinagari, Modi and Tulu-Tigalari.
IIT Roorkee has gone further down one script. Its researchers have developed MoScNet, the first AI model able to transliterate the ancient Modi script into Devanagari, under Professor Sparsh Mittal with students Harshal and Tanvi of COEP Pune and Onkar of the Vishwakarma Institute, as part of the AI ancient manuscripts effort. The project, called Historic Scripts to Modern Vision, also includes a dataset named MoDeTrans with more than 2,000 verified images of original Modi manuscripts and matching Devanagari transliterations. An estimated 40 million documents in Modi script exist across India, including land deeds, scientific manuscripts and Ayurveda texts, and very few people alive today can read the script. The model uses a vision-language architecture and has been open-sourced on Hugging Face with the dataset.
AI ancient manuscripts: what readers will be able to do
The AI ancient manuscripts mission builds on older work. The National Mission for Manuscripts, launched in 2003, has spent two decades cataloguing and preserving manuscripts across India, and Gyan Bharatam takes that work into digitisation and AI. The difference is access: a student or researcher will be able to ask the repository a question in an Indian language and get an answer drawn from the manuscripts, with the page cited. The ministry has not given a date for the full repository to open, and the technology partner for the platform is still being engaged. The portal is already live, the funding is sanctioned, and the first tools are listed, so the AI ancient manuscripts work of turning more than one crore manuscripts into searchable text has a start date even if its finish is years away.