By Lukie Pieterse, Potato News Today
Learn how this innovative tool enhances crop improvement to support farmers and global food security.
Buba is an innovative AI-powered platform developed as part of the International Potato Center’s (CIP) AI-griculture Challenge Hackathon in 2024. Announced as the winner in the Crop Improvement Science Goal category on May 23, 2025, Buba addresses key challenges in potato breeding by leveraging artificial intelligence to enhance the selection and optimization of potato varieties.
The hackathon, which attracted over 60 teams worldwide, focused on using AI to tackle global food security issues, with Buba standing out for its potential to accelerate resilient potato development amid climate change and disease pressures. As a winning entry, the team behind Buba received $4,000 in prize money, along with opportunities for collaboration, scaling, and implementation through CIP’s network.
Potato breeding is traditionally a lengthy, resource-intensive process involving manual trait evaluation, cross-breeding, and field testing over multiple seasons. Buba streamlines this by integrating predictive modeling and data-driven insights, making it a valuable tool for breeders, researchers, and agronomists working on root and tuber crops.
Below is a comprehensive breakdown of what Buba offers to potato breeders, based on its core design and objectives.
Key Features and Capabilities
Buba’s architecture centers on machine learning models trained on potato genetic data, environmental variables, and historical breeding outcomes. It optimizes trait combinations to predict and prioritize varieties that perform well under future conditions. Here’s a detailed look at its offerings:
| Feature | Description | Benefits for Potato Breeders |
|---|---|---|
| Trait Optimization Engine | Uses AI algorithms to analyze and recommend optimal combinations of potato traits (e.g., yield potential, tuber size, skin quality, and nutritional content) based on user-defined goals. | Reduces trial-and-error in cross-breeding by simulating thousands of combinations virtually, shortening breeding cycles from years to months. |
| Climate Change Adaptation Modeling | Incorporates climate projections (e.g., temperature shifts, drought patterns, and precipitation changes) to evaluate how traits will perform in future scenarios, drawing from geospatial and historical weather data. | Enables breeders to develop varieties resilient to extreme weather, critical for regions like sub-Saharan Africa and the Andes where potatoes are staple crops but vulnerable to warming trends. |
| Disease Resistance Prediction | Integrates data on pathogens (e.g., late blight Phytophthora infestans, common scab) to forecast resistance levels in potential hybrids, factoring in genetic markers and environmental triggers. | Minimizes losses from diseases that cause up to 60% yield reductions globally; allows proactive selection of blight-resistant lines without extensive field exposure. |
| Predictive Performance Simulation | Runs scenario-based simulations to estimate field performance, including yield forecasts and adaptability to specific agroecological zones. | Supports data-informed decisions for germplasm selection, aligning with CIP’s Genebank integration goals for efficient seed requests and trials. |
| User-Friendly Interface and API Integration | Features a dashboard for visualizing recommendations, with potential API hooks to connect with CIP’s databases (e.g., potato and sweetpotato catalogues) for seamless data flow. | Democratizes access for breeders in low-resource settings, enabling collaboration across teams without advanced coding skills. |
How Buba Supports Potato Breeders in Practice
- Accelerated Breeding Pipelines: Traditional potato breeding can take 10–15 years to release a new variety. Buba’s AI models, trained on datasets like those from CIP’s breeding programs, can identify promising candidates early, potentially cutting this timeline by 30–50% through targeted predictions.
- Enhanced Resilience and Sustainability: By prioritizing traits for climate and disease resilience, Buba aligns with global goals like the UN’s Sustainable Development Goal 2 (Zero Hunger) and CIP’s 2030 strategy for biodiversity and nutrition. For instance, it could help breed varieties with higher drought tolerance or reduced fungicide needs, lowering costs for smallholder farmers who produce 95% of potatoes in developing countries.
- Data Completion and Standardization: Building on the hackathon’s Challenge 3 (Enhancing Germplasm Selection), Buba likely uses natural language processing (NLP) to fill gaps in incomplete datasets from research papers and reports, ensuring breeders have comprehensive trait profiles.
- Scalability and Customization: As a hackathon prototype, Buba is designed for expansion—breeders can input local data (e.g., soil types or regional pests) to tailor recommendations, making it adaptable for diverse contexts like highland Peru or East African farms.
Development and Future Potential
The Buba team, consisting of Shadrack Odikara and Meshak Emakunat, collaborated with CIP mentors and judges, including experts in crop improvement and AI. The tool draws from the hackathon’s emphasis on completing potato germplasm datasets and building predictive models for traits like yield and nutritional quality. Post-win, CIP has indicated opportunities for implementation, such as integration into their Genebank request system or field trials in partner countries.
Looking ahead, Buba could evolve into a broader platform, potentially incorporating real-time drone imagery or genomic sequencing for even more precise predictions. CIP’s ongoing commitment to AI in agriculture suggests Buba may see pilots in 2026, contributing to resilient potato varieties that support 1.5 billion people reliant on the crop.
For more details or to connect with the team, breeders can reach out via CIP’s innovation channels at cipotato.org. This tool exemplifies how AI is transforming crop improvement from an art to a science, promising higher yields and food security in a changing world.
References for further information: