Roots and Tubers to Go: CGIAR / CIP

The Arizona State University Artificial Intelligence Cloud Innovation Center, powered by AWS collaborated with the Consultative Group on International Agricultural Research (CGIAR) and the International Potato Center (CIP) to develop an interactive, generative AI guide to support requests from Researchers seeking potato and sweet potato accessions (seeds and other seed-like matter). CIP is the leading seedbank (genebank) for potatoes and sweet potatoes, and serves as a central location receiving requests for germplasm from the global research community and local farmers. This post reviews the creation of a GenAI bot to enable the CIP team to quickly receive the necessary information to accurately respond to the requesting researcher, reducing the standard 3-4 month back and forth process to days. 

To align on this solution, CGIAR and CIP utilized Amazon’s approach to innovation, Working Backwards. This process helps customers go from a pain point to a solution. Through this mechanism, customers align on a priority project and then proceed to the development of a proof of concept and ultimate production deployment.

Researchers encounter difficulties searching for accessions using online databases with limited search functionality and incomplete data availability. These existing databases are difficult to navigate, and produce too many results to be useful. As a result, researchers email the CIP team directly to find the right accessions needed to fulfill their project needs. These researchers are often not providing all necessary accession properties to the CIP team to accelerate the search process. Due to the lack of comprehensive requirements provided, this resulted in months of back-and-forth emailing with the researchers to narrow the request and assure the CIP genebank was providing the right plant products. This caused delays in the distribution of potato and sweet potato genetic resources for breeding, training and research, and slowed progress in the research and development of these important crops. These inefficient processes on the CIP side also resulted in hours being spent working on back-and-forth communications as opposed to direct mission focused work. 

The CGIAR project leveraged the power of AWS to develop a robust and scalable backend architecture for the chatbot. By utilizing services such as Amazon ECR, Amazon ECS, and Application Load Balancer (ALB), the team was able to create a highly available and efficient deployment environment.This project was developed using AWS and the Streamlit framework.The backend flow of the chatbot involves triggering the Bedrock API when a user enters a query, which then interacts with the LLM (Claude Sonnet) to provide a response and collect the user's required traits. When a document is uploaded by a researcher to the chat, it is stored in a dedicated S3 Bucket for the respective session. After gathering all the necessary information, the chatbot sends an email to the user using the AWS SES service, providing the S3 path to their documents and the collected traits.

The CGIAR chatbot solution is a testament to the power of collaboration and the transformative potential of AI-driven technologies. By addressing the longstanding challenges faced by the scientific community in accessing and requesting potato and sweet potato germplasm, this project has paved the way for a more efficient and user-friendly approach to agricultural research and seed bank management.

“We encountered a recurring issue where user requests for germplasm recommendations were sent to various departments across the entire center instead of using the designated online system. This often led to lengthy and stalled email exchanges, causing frustration. To address this, we sought a solution leveraging artificial intelligence for more immediate interaction. By working backwards, the ASU AI CIC helped us understand users' most urgent problems and focus on solving them. By the end of the workshop, we had a clear idea of what needed to be improved in the request system. Together with AWS, we co-created a chatbot to quickly gather information and provide germplasm recommendations, reducing processing time from months to minutes. Following the workshop AWS then developed a pilot chatbot in just a few weeks. Looking ahead, we plan to use AI to enhance the quality of the querying, matching the query result to the available germplasm in the CIP Genebank and integrating it with the requesting system. We aim to also improve the information to users  by integrating CIP databases with external online information”

-Dr. Bettina Heider I Genetic Resources Specialist, Genebank International Potato Center

Through this innovative chatbot, the CGIAR and CIP teams have demonstrated how cutting-edge technologies can be leveraged to revolutionize the way organizations manage and utilize plant genetic resources. As the solution continues to evolve and expand its capabilities, the impact on the broader agricultural landscape is poised to be profound, driving advancements in crop development, food security, and sustainable agriculture.