Overview
Kenyan SACCOs (savings and credit cooperatives) manage complex member portfolios but often lack accessible tooling for non-technical staff. This project bridges that gap with a conversational AI interface backed by 14 MCP tools covering the full financial operations surface.
Staff can ask questions like "What is the PAR for the month?" or "Show me members with overdue loans" and receive structured, accurate answers without writing SQL or navigating complex dashboards.
Key Capabilities
- Natural language queries in English or Swahili over member, loan, and financial data.
- 14 MCP tools covering PAR analysis, liquidity metrics, portfolio health, and member management.
- Role-based access control and admin controls for sensitive financial operations.
- File upload support for bulk data ingestion.
- Comprehensive financial metrics including portfolio at risk, liquidity ratios, and loan performance.
Stack
- React (frontend)
- FastAPI (backend)
- Groq Llama (LLM)
- Model Context Protocol (MCP)