Large-scale LLM applications must evolve their architectures to handle uncertainty and capability gaps. Our aim is to shift from an orchestrator-based architecture to a Model Context Protocol (MCP) agentic system for Fanar, the comprehensive Islamic AI assistant.

Java client for the Fanar AI platform — Qatar's Arabic-centric multimodal AI. Type-safe access to chat, vision, speech, translation, poetry, and Islamic RAG APIs.