where AI acceleratesSCIENCE

*/01*

THE MISSION

Uniting the world's brightest minds to build intelligent agents that don't just analyse data but actively collaborate in the research process—accelerating breakthroughs in humanity's most critical fields.

*/02*

THE ECOSYSTEM

Our open platform spans the entire AI-for-science development lifecycle—fostering collaborative events, open-source infrastructure, and shared evaluation standards.

Connect

ai4.science brings together researchers, developers, and AI experts through open collaborative events. Our hackathons foster innovation in AI-human scientific collaboration, while future markathons will focus on creating shared benchmarks for open AI evaluation in scientific domains.

Build

mcp.science provides the open-source infrastructure and tools researchers need. Our Model Context Protocol servers offer the technical foundation for building AI agents that truly understand and contribute to scientific research workflows.

Test

bench.science establishes open evaluation frameworks for AI capabilities in scientific research. Our community-driven benchmarks ensure AI systems can reason effectively about complex scientific problems and contribute meaningfully to collaborative research outcomes.

*/04*

Innovation Showcase

Inaugural Event: Stanford Quantum Science Hackathon

View Event Details

Our first event at Stanford University brought together researchers from quantum science and engineering to build applications where AI agents contribute meaningfully to real scientific tasks—from computation and simulation to experimental control and ideation.

Quantum Circuit Design

Closed-loop planar circuit QED design with LLM - revolutionizing quantum circuit optimization through AI-driven design iterations.

Matthew Chalk, Eesh Gupta, Kaveh Pezeshki, Yueheng Shi, Wendy Wan

Quantum Hardware

AI-powered engine for designing superconducting quantum chips, automating the entire process from layout to electromagnetic simulations.

Sebastien Boussard, Gabriel Dupuis, Sarah Dweik, Paul Goldschmidt

AI Infrastructure

A framework that allows AI agents to learn from mistakes and autonomously expand their capabilities when working with scientific Python libraries.

Wanda Hou, Hongye Hu, Miao Li

Astrophysics

Simplifies complex simulations like black hole collisions by automating the entire workflow from setup to gravitational wave visualization.

Jiani Fei, Jiayi Hu, Nan Sheng, Shuo Xin, Rong Zhang

Information Theory

AI agent that creates and tests error correction codes, implementing complex algorithms for data transmission reliability.

Henry Hunt

Quantum Chemistry

Bridges AI agents with PySCF quantum chemistry framework, enabling natural language requests for complex computational chemistry.

Andrew Jenkins, Lixin Lu

Hackathon · Shanghai

View Event Details

Nearly 200 researchers, engineers, and builders gathered at Shanghai Jiao Tong University’s USC Institute to advance agentic AI for science. Teams delivered 21 prototypes across lab automation, materials discovery, data analysis, and workflow security.

Economics

Combines data exploration, literature grounding, and econometric modeling to help researchers formulate and validate causal hypotheses automatically.

Wang Zhiyuan, Zhang Han

Quantum Materials

Integrates high-throughput screening, first-principles simulation, and GANs to accelerate materials discovery with MCP agents.

Jiang Haoyan, Li Zhenghong, Wang Qixiang

Hackathon · Beijing

View Event Details

At Tsinghua University’s Physics Building, teams pushed MCP-powered human–AI collaboration across quantum computing, advanced materials, and energy systems—showcasing how quickly agentic workflows translate into working tools.

Quantum Computing

Provides an intelligent, automated compilation pipeline for neutral atom quantum computers, covering the full stack of compilation and optimization.

Zhang Tao, Nie Xiaotian, Ni Zhongyi, Zhou Tiangang

Battery Systems

Builds an LLM-driven battery management system that analyzes pack health and safety states with MCP-connected diagnostics.

Chen Wenjie, Lin Zichang, Shi Xiaodong