Crystalyse: a multi-tool agent for materials design

TL;DR

Crystalyse is an open, multi-tool AI agent that streamlines inorganic materials design by integrating compositional screening, structure generation, and machine learning evaluation, with modes balancing exploration and validation.

Contribution

It introduces a provenance-enforced, multi-mode AI agent for materials design, enabling rapid exploration, adaptive routing, and comprehensive validation, with open source tools for reproducibility.

Findings

Provenance enforcement reduced hallucinations from 57% to 14%.

Agent successfully designed new inorganic materials in diverse applications.

Open source code facilitates plug-and-play use and further development.

Abstract

We present Crystalyse, an open, provenance-enforced scientific agent for computational materials design of inorganic crystals that orchestrates tools for compositional screening, crystal structure generation, and machine-learning force-field evaluation. Crystalyse offers three operating modes to trade exploration speed against validation depth: creative (rapid query), adaptive (context-aware routing) and rigorous (comprehensive checks). We release the underlying source code and evaluation scripts to enable plug-and-play use and development. In demonstrations on quaternary oxide exploration, sodium-ion cathode design, and lead-free indoor photovoltaic candidate generation, the agent integrates chemical compound generation with fast stability and property filters. Under adversarial testing, provenance enforcement eliminated material-property hallucinations (a broad adversarial suite pass rate reached 86% from a 57% baseline). Crystalyse provides an agentic artificial intelligence system that can complement existing materials design pipelines, assisting in hypothesis generation while preserving transparency and reproducibility.