This repository curates resources for Agentic Deep Research, targeting researchers and developers interested in AI-powered autonomous research. It provides a structured overview of industry products, open-source tools, recent papers, and benchmarks in the field, aiming to be a comprehensive guide to this rapidly evolving domain.

How It Works

The project acts as a curated knowledge base, categorizing advancements in agentic deep research. It highlights key trends such as the shift from traditional web search to agent-driven deep analysis, the integration of Retrieval-Augmented Generation (RAG) with reasoning capabilities, and the development of multimodal research agents. The inclusion of a detailed table of recent research papers with their methodologies, models, and datasets offers a structured view of the state-of-the-art.

Quick Start & Requirements

This repository is a collection of links and information, not a runnable software package. No installation or specific requirements are listed for the repository itself.

Highlighted Details

• Features a comprehensive table of recent research papers (2025) detailing agent architectures, base models, optimization techniques, and evaluation datasets.
• Lists industry-leading products and open-source implementations from major tech companies and research labs.
• Includes links to relevant benchmarks for evaluating agentic search and reasoning capabilities.
• Provides citation information for two key position papers on agentic deep research and RAG-reasoning systems.

Maintenance & Community

The repository is actively maintained, with recent updates listed throughout 2025. Contributions are welcomed via GitHub issues or pull requests for new papers, tools, or products. Citation information is provided for the project's own research papers.

Licensing & Compatibility

The repository itself does not appear to have a specific license mentioned. The content is a curated list of external resources, whose individual licenses would need to be checked separately.

Limitations & Caveats

This repository is a curated list and does not provide executable code or direct access to the tools and research it lists. The information is presented as a snapshot of ongoing research and product development, with many entries dated in the future (2025), indicating a forward-looking but not yet fully realized landscape.