Eneos AI system 'to speed up' materials R&D work

Tokyo – Eneos Materials has developed an AI system designed to accelerate the creation of new materials through combining the company’s in-house “knowledge graph” with “AI agent technology”.

The “new-materials theme creation AI agent system” entered internal verification in October, working on new R&D themes from the perspective of “materials x applications x functions” based on the needs of the market.

The tool has "already shown promise in early testing," said Eneos 28 Nov.

According to Eneos Materials, internal proof-of-concept trials demonstrated that the system can make first-stage screening of candidate materials around 75% more efficient compared to previous in-house processes.

Eneos said its knowledge graph - a knowledge structure that systematically links events, concepts and relationships in documents and data – was built by analysing internal research reports and technical documents.

It consists of around 30,000 entities and 80,000 relationships, mapping concepts, materials and links between them.

The generative AI engine uses “a multi-agent system”, in which several AI agents collaborate, evaluate one another’s outputs and refine suggestions.

Eneos Materials said the system provides “a data-driven foundation” for identifying material combinations that conventional generative AI would not reach.

This, it added, will help shorten development lead times and improve overall efficiency in the R&D process.