Marking a major step in taking AI from boardrooms to Bharat, the collaboration brings Vision AI to the grassroots economy, translating digital promise to actual rural impact
New Delhi, The Tobacco Board of India, Guntur, has partnered with Constems AI Pvt. Ltd., a deep-tech Vision AI company, to introduce AI-powered grading of Flue-Cured Virginia (FCV) tobacco bales across a few auction platforms in Andhra Pradesh. The collaboration will use computer vision and deep learning to make tobacco grading faster, more consistent and more transparent.
Within the Tobacco Board’s auction ecosystem, FCV tobacco grading is currently carried out manually by qualified classifiers. While the process is backed by domain expertise, it remains dependent on human judgement and can be susceptible to subjectivity and inconsistency. Since the grade assigned to a tobacco bale directly influences its market value, a standardized and technology-supported grading framework has become increasingly important.
The implementation is expected to create a uniform AI-assisted grading system for farmers, buyers, exporters, regulators and the wider tobacco industry.
Constems AI’s CAInatics, its proprietary Vision AI engine, works in conjunction with 3N, the company’s proprietary Large Vision Modal Model (LVMM), to analyse images of tobacco bales coming from different sources and generate grading output. The results are delivered instantly on the grader’s mobile application, enabling faster and more consistent decision-making at the auction platform.
The solution provides individual parameter scores along with an overall combined grade based on key quality attributes such as colour, texture/grain and visible leaf condition. Since the Tobacco Board’s grade framework is already rooted in visual indicators, the process is naturally suited for AI-led assessment that can read, compare and classify tobacco bales with greater consistency.
For farmers, grading directly shapes value realization. A more consistent and transparent grading system can strengthen confidence that the product is being assessed against defined benchmarks. For buyers and auction officials, it can improve traceability, reduce variation and support faster assessment across large volumes of tobacco bales.
Amit Srivastava, Co-Founder and CTO, Constems AI, said, “Every tobacco leaf carries visible quality signals, from colour and grain to blemish and injury. The challenge is to read these signals consistently, at scale and in real time, especially in an auction environment where grading directly influences farmer value. Through this collaboration with the Tobacco Board of India, we are bringing Vision AI into one of the most critical points of the agricultural value chain. This is India-built deep technology working inside a real marketplace, helping create greater transparency, consistency and confidence for farmers, buyers and the wider tobacco ecosystem. For us, this is also part of a larger vision:
to build AI that does not remain confined to controlled environments, but becomes a cognitive vision layer for the real world, helping industries, institutions and communities see better, decide faster and act with greater confidence.”
Secretary, Tobacco Board of India, said, “The Tobacco Board is conducting e-auctions for the marketing of FCV tobacco, and tobacco classification is the only manual operation in the auction process. AI-based grading of tobacco would reduce manual intervention and automate the only existing manual operation in the e-auction system. Successful implementation of AI-assisted grading is expected to enhance transparency, consistency, and efficiency in tobacco classification at auction platforms and enhance the confidence of both growers and buyers.”
The collaboration marks an important step in the use of Vision AI for agricultural grading and rural commerce. It demonstrates how advanced technology can support not just enterprise efficiency, but also the systems that influence market access, value discovery and farmer livelihoods.
For Constems AI, the mandate is another step towards its larger ambition of building the cognitive vision layer powering the next era of execution across enterprises, infrastructure and society.
