At the 2026 Tencent Global Digital Ecosystem Summit, Wenge AI demonstrated a significant advancement in the field of AI for Science (AI4S) by showcasing an AI-assisted hydrogen-production catalyst design workflow. Powered by ScienceOra, Wenge AI's intelligent research decision platform, the demonstration illustrated how advanced artificial intelligence can bridge the gap between academic literature, material formulation, experimental planning, and industrial application.

The demonstration focused on a highly complex materials science challenge: designing a high-entropy alloy catalyst for the alkaline hydrogen evolution reaction. By connecting scientific models, professional literature, and research data, the platform aims to help materials researchers translate scientific evidence into actionable research plans.

Bridging Literature and Material Formulation

The design of advanced catalysts typically requires navigating vast amounts of scientific literature to identify promising material compositions. In the demonstration, ScienceOra addressed the need for high activity and durability in alkaline hydrogen evolution reactions by searching professional literature and drawing on published research methods and elemental properties.

Rather than acting as a simple generator of material formulations, the platform connects documentary evidence, materials knowledge, and specific design objectives. This approach supports a more targeted analysis of candidate solutions. Key aspects of this phase include:

  • Drawing on published research methods and elemental properties to support candidate designs.
  • Cross-checking optimization proposals against relevant literature to provide researchers with verifiable references.
  • Ensuring that human researchers remain responsible for assessing scientific validity and determining subsequent research directions.

From Lab Bench to Industrial Scale

Beyond initial candidate design, the workflow demonstrated by Wenge AI extends into experimental planning and industrial decision support. This comprehensive approach ensures that theoretical designs can be successfully validated in the laboratory and eventually scaled for industrial use.

For experimental planning, ScienceOra assists in developing research protocols around candidate designs. The platform helps researchers organize experimental variables and produces testing standard operating procedures (SOPs) and related reference information for subsequent validation.

For industrial applications, the platform supports comparative analysis across multiple critical dimensions:

  • The feasibility of scaled hydrogen production.
  • Performance at industrially relevant current densities.
  • Durability and raw-material availability.
  • Compatibility with mass-production synthesis processes.

These candidate designs, experimental plans, process recommendations, and references are consolidated into a structured research report. This allows research and development teams to assess scientific evidence alongside engineering constraints, though experimental testing and scientific validation remain essential parts of the materials research process.

Real-World Impact and Ecosystem Integration

The practical utility of this workflow is supported by a previously disclosed industry-research collaboration on hydrogen-production catalyst design. In that collaboration, ScienceOra supported literature analysis and research-plan development, resulting in a reported 38% improvement in catalyst activity. Furthermore, the collaboration reduced the design stage from several months to 30 minutes, illustrating the potential of domain-specific AI to connect scientific literature, research design, and engineering requirements.

To make these capabilities more accessible, ScienceOra was recently made available on Tencent WorkBuddy. The initial offering includes more than 30 research expert agents and skills covering literature analysis, research hypotheses, plan design, modeling and simulation, validation analysis, and research communication.

Wenge AI (1956.HK), founded by a team of AI scientists from the Chinese Academy of Sciences, continues to focus on advancing AI from technological capabilities to enterprise-level applications. The demonstration at the summit highlights how AI4S is moving beyond general information processing toward support for specialized research workflows and industrial R&D decision-making.