Towards Inclusive Financial Guidance: A Multilingual Nigerian-Aware Chatbot Built with LangChain and NLLB-200

DeepLearningIndaba 2025 Workshop AIBF Submission7 Authors

Keywords: Multilingual, Financial, Chatbot, LangChain

TL;DR: **TL;DR** We present a multilingual AI financial advisor chatbot that combines LLMs, document retrieval, translation, and Nigerian voice synthesis to deliver accessible, culturally relevant financial guidance across low-resource African languages.

Abstract: This paper introduces a multilingual AI-powered financial advisor and planner chatbot designed specifically for Nigeria’s linguistically diverse population. Leveraging LangChain's orchestration capabilities, OpenAI's GPT-4o for advanced natural language understanding, Pinecone for semantic document retrieval, and Meta's NLLB-200 model for high-quality translations, the system delivers personalized financial advice in five languages: English, Nigerian Pidgin, Yoruba, Igbo, and Hausa. The chatbot architecture incorporates context-preserving memory via LangChain’s ConversationBufferMemory, dynamic prompt engineering using ChatPromptTemplate, and Pinecone-powered RAG (Retrieval-Augmented Generation) to ensure responses are both accurate and evidence-based. For accessibility and engagement, localized text-to-speech functionality is added via Spitch TTS, enabling auditory feedback in native Nigerian accents. This solution bridges language and digital literacy gaps, enabling users, especially those in underserved communities, to access budgeting, investment, and debt management support in their preferred language. Our prototype demonstrates a practical deployment path for inclusive, AI-driven financial services in emerging markets, offering a blueprint for future development across multilingual, low-resource settings.

Submission Number: 7