Chatbot for Food Policy Analysis
Year
Technologies
Navigating complex food policy documents can be challenging. GAIN contracted rowsquared to develop a custom AI chatbot for Bangladesh that simplifies the analysis and improves the accessibility of those documents for a broad range of stakeholders, from government officials to technical experts and the general public.
We built an AI chatbot that processes these documents by storing them in a searchable database. End users can submit simple, natural language queries, and the chatbot retrieves relevant information from the documents, providing clear and concise responses to complex policy topics.
The system leverages a Retrieval-Augmented Generation (RAG) architecture, combining embedding technologies and large language models (LLMs) to efficiently process user queries and retrieve accurate information.
A key challenge was accommodating end users with limited experience in AI and chatbots. To address this, our team focused on creating a user-friendly interface, seamlessly integrated into GAIN’s website, that made advanced functionalities easy to navigate.
The chatbot supports comparisons of multiple documents, aiding in the analysis of policy coherence and discrepancies.
Comprehensive technical documentation was also provided, facilitating future adaptation and scaling of the chatbot to additional countries.
The positive feedback on this pilot has led GAIN to explore the possibility of scaling the solution to other countries, demonstrating the potential for broader application.
The chatbot can be accessed at https://policybot.gainhealth.org.
