Materealize: a multi-agent deliberation system for end-to-end material design and synthesis

Seongmin Kim, Jaehwan Choi, Kunik Jang, Junkil Park, Varinia Bernales, Alan Aspuru-Guzik, Yousung Jung

Submission Type: I want my submission to be considered for both oral and poster presentation.

Keywords: Agentic AI, Multi-Agent Debate, End-to-End Material Discovery, Synthesis planning, Inorganic Crystals

TL;DR: Materealize is a multi-agent framework for end-to-end materials discovery. Compared with tool-only workflows, it adds mechanism-level synthesis guidance, reaches 91.4% top-5 precursor prediction accuracy, and expands realizable design space by 8.3%.

Confirmation Of Submission Requirements: I submit an abstract. It uses the template provided on the submission page and is no longer than 2 pages.

Submission Number: 364