Recent advancements in large reasoning models (LRMs) have been highlighted through the introduction of Search-o1, an agentic search-enhanced framework designed to improve the integration of external knowledge into LRM processes. This new framework aims to address the issue of knowledge insufficiency that often plagues standard LLMs. Notably, Diffbot has developed a natural language interface that utilizes its knowledge graph technology, promising enhanced performance over existing search engines such as Google Gemini and ChatGPT. The Search-o1 model incorporates a reason-in-documents module and a multi-agent system called MAIN-RAG, which enables LLMs to collaboratively filter and rank retrieved documents, thereby increasing the reliability of answers without additional training. These developments are part of a broader trend in artificial intelligence aimed at creating more intelligent applications through structured metadata and optimized retrieval operations.