Bohrium + SciMaster AI Agent Competition
Making Creation Sustainable
This competition aims to promote the development of research AI agent technology, encourage innovative applications, and showcase excellent works. Welcome all individuals and teams interested in research AI agent development to participate!
AI Agents Are Reshaping Research Paradigms
Society supports scientific research because exploration expands the boundaries of human knowledge, bringing technological innovation that benefits humanity. AI Agents are becoming the transformative engine of intelligent research.
Unlike traditional tools that provide point-based assistance, research AI agents can drive and integrate literature reading, model computation, experiments, and other workflows, truly achieving the transition from 'can do' to 'know how to do'. Scientists will have cross-domain, cross-system AI research partners to help integrate knowledge and solve problems; repetitive work will be completely freed, and innovative ideas will be proposed, verified, and iterated faster.
Research AI agents may introduce new evaluation and incentive models for intellectual achievements. The basic unit of work output gradually transforms from static papers to dynamic, executable agents. These agents form a real-time flowing value network through high-frequency mutual invocation.
Cross-domain Collaboration
AI research partners help integrate knowledge and solve interdisciplinary challenges
Efficiency Boost
Repetitive work is freed, innovative ideas are verified and iterated faster
Value Network
Agents invoke each other, forming a real-time flowing value network
New Evaluation
From static papers to dynamic agents, evaluation becomes more real-time and objective
Two Tracks, Independent Evaluation
Choose the most suitable track based on your expertise and application scenarios
Deep Research Track
Domain-specific Applications
Focus on specific disciplines such as materials, chemistry, biology, and simulation to solve professional research problems
General Research Track
Cross-disciplinary Scenarios
Address general research scenarios across disciplines, including literature reading, data analysis, and experiment management
Total Prize Pool ¥1,000,000
Two independent tracks with generous prizes and continuous support for outstanding projects
Data-Driven + Expert Review
Unique dual evaluation system: real user feedback meets professional expert judgment
Platform Revenue Performance
Your work must not only tell a good story but also win users over. We evaluate the actual value of works through real platform data.
Professional Evaluation
A panel of experts from research and technology fields conducts professional evaluation from multiple dimensions.
The degree of innovation in technical approach, application scenarios, or solutions
Product maturity, stability, and user experience
Practicality and promotion potential in real research scenarios
Why This Evaluation Method?
We believe good research tools must not only be innovative but also truly help researchers solve problems. 50% objective data ensures works withstand market validation, while 50% expert review ensures technical depth and innovative value are recognized.
Open, Fair, Innovative
We welcome all individuals and teams interested in research agent development
Eligibility
Open to researchers and developers worldwide, with no restrictions on location or professional background
Originality
Submissions must be original work by participants, without plagiarism or infringement of intellectual property rights
Content Standards
Submissions must comply with relevant laws and regulations, and must not contain any illegal, inappropriate content (such as violence, pornography, discrimination, etc.)
Please refer to theofficial competition documentation for detailed rules. Contact the organizing committee if you have any questions
Competition Timeline
From ideation to implementation, supporting teams throughout the journey
Launch
Phase 1: Creative design, development attempts, and online submission. Demonstrate the core value of research AI agents.
Trials
Conducted as an online defense. Select 25 teams from 100+ entries to advance to finals.
Finals
Top 30 teams compete. Two tracks evaluated independently across multiple dimensions, including innovation, completion, and practical value.
Award Ceremony
Final rankings will be announced and prizes awarded at the Beijing Science Intelligence Summit.
Making Creation Sustainable
Competitions may have winners and losers, but good ideas should not be abandoned due to time constraints. Through this competition, we hope to incubate, grow, and sustain more excellent research AI agent projects.
Promote the transformation of research evaluation systems, making the value of intellectual achievements more real-time, objective, and accurate
200,000 yuan Incentive program fund pool, giving excellent projects a second chance and supporting long-term development
