Multi-agent AI research assistant orchestrates specialized agents for academic paper discovery, retrieval, and synthesis using arXiv, Semantic Scholar, and OpenAlex APIs. A FastAPI backend routes user queries through an Orchestrator that sequences Search, Filter, Reader, Retrieval, Summarizer, Analysis, and Report agents, each handling distinct tasks in the research pipeline. Text chunking, semantic embeddings via FAISS vector database, and Ollama LLM inference enable intelligent document processing and synthesis without external LLM dependencies. Fork this architecture on Diagrams.so to customize agent workflows, swap vector stores, or integrate proprietary research APIs for your domain. The modular agent design supports easy addition of new specialized agents for domain-specific research tasks.