In a significant stride toward bridging the gap between livestock science and practical animal health decision-making, the International Livestock Research Institute (ILRI) has developed an AI-powered chatbot that delivers instant, evidence-based answers to animal health questions.
The tool—accessible at animalanswers.ilri.org—was developed by ILRI’s Data and Research Methods Unit (DRMU) and the CGIAR Digital Transformation Accelerator (DTA) in collaboration with Kenyan tech startup Fahamu AI.
The chatbot is part of ILRI’s broader effort to digitalize livestock research and knowledge dissemination and to make complex scientific outputs more usable for frontline livestock producers, extension workers, veterinarians, and policymakers.
By tapping into CGIAR’s open-access repository of scientific publications housed in CGSpace, the tool enables users to ask questions in plain language and receive rapid responses grounded in peer-reviewed research.
Livestock diseases remain a major constraint on productivity and livelihoods across Africa and other low-resource regions, particularly for camel, cattle, and small ruminant herders who may lack timely access to credible animal health advice.
Traditional research outputs—scientific papers, reports, and datasets—are not always accessible or actionable for field practitioners who need quick, practical information without wading through extensive literature.
The chatbot is designed to help overcome these barriers by extracting and synthesizing relevant knowledge from CGIAR’s extensive research corpus.
How the chatbot works
The interface of the AI-powered chatbot is simple and intuitive: users type a question into the main text box and click “Ask.” The system then generates an answer that includes clickable citations linking back to the source materials, or “chunks,” drawn from scientific publications.
This transparency allows users to trace the evidence behind each response and explore the full references at their own pace.
A built-in feedback mechanism lets users rate responses on accuracy, completeness, and helpfulness, providing developers with real-time insights into performance and reliability.
Additionally, the chatbot features a history tool that enables users to revisit past questions and responses, supporting learning and continuity over time.
According to Jean-Baka Domelevo Entfellner, Head of Data and Research Methods at ILRI, the initiative aims “to make our research outputs truly actionable by letting anyone ask questions in plain language and receive reliable, sourced answers instantly.” The collaborative effort also included technical support from Alan Orth, a systems specialist within DRMU.
Pilot phase and user feedback
The chatbot is currently in an early pilot phase, with all major functionalities operational but still undergoing refinement based on user feedback.
A hands-on workshop held at the ILRI campus in Nairobi brought together researchers, developers, and prospective users to test the tool and identify areas for improvement.
While participants praised the chatbot’s potential to democratize access to scientific knowledge, they also highlighted several technical challenges.
One issue was the inconsistent clarity of source attribution: in some cases, cited sources appeared simply as “document excerpt” without clear bibliographic details, reducing confidence in traceability.
Users also noted that relevance scoring for cited document chunks sometimes ranked loosely related sources too highly, and that occasional unintelligible text cropped up when the system was tested with off-topic queries.
Suggestions from the workshop’s participants are now informing a roadmap for enhancing the tool. Recurring recommendations included replacing the fixed limit of five document chunks per response with a relevance threshold that surfaces only the most meaningful sources.
Users also expressed interest in features such as personalized login capabilities, multi-turn conversation threads, and selectable user personas—such as “farmer” or “scientist”—to tailor responses’ tone and depth.
Broader vision and the road ahead
The AI-powered chatbot exemplifies CGIAR’s push to harness digital innovation for agricultural transformation, aligning with the Digital Transformation Accelerator’s mission to integrate AI tools that democratize access to research and improve decision-making across agri-food systems.
For ILRI, the chatbot marks a milestone in using digital tools to scale the impact of livestock research. The DRMU team envisions this first tool as the foundation of a broader suite of domain-specific AI applications, covering areas such as livestock genetics, nutrition, and climate adaptation.
“We’re starting small, but the vision is big,” Domelevo Entfellner said, underscoring the institute’s plans to expand and diversify its digital knowledge offerings.
Looking ahead, planned improvements include giving users more control over answer length, exploring the integration of generative AI content alongside corpus-based retrieval, and experimenting with alternative generative AI APIs to optimize speed and accuracy.
There is also interest in enhancing back-end performance by integrating GPU-based hosting to boost processing efficiency.
The chatbot is publicly available at animalanswers.ilri.org, and ILRI invites users to try it out, explore the references, and share their experiences to help shape its next iteration.
Source : Farmers Review Africa
