By minimizing the technical barriers traditionally associated with computational chemistry, El Agente is a step toward more inclusive, accessible, and scalable scientific research worldwide.
The Matter Lab at the University of Toronto is pleased to introduce El Agente, a large language model (LLM)-powered multi-agent system designed to make computational chemistry more accessible. By combining natural language understanding with dynamic tool orchestration, El Agente enables users—from expert researchers to newcomers and aficionados—to perform complex quantum chemistry tasks to evaluate molecular properties and determine their behaviour.
Computational chemistry has become indispensable in modern research. It continues to drive innovation in drug discovery, materials science, and catalysis. But despite its transformative potential, widespread adoption remains limited by steep learning curves, complex software interfaces, and the expertise required to manage errors and interpret results. Existing workflow systems partially address these challenges but often lack the flexibility needed for novel or domain-specific tasks, limiting their utility for non-experts.
“LLMs have great potential to transform the way we do science,” says Alán Aspuru-Guzik, principal investigator at the Matter Lab, director of the Acceleration Consortium and Senior Director of Quantum Chemistry at NVIDIA. “El Agente democratizes access to computational chemistry, an important step that will lead to even greater innovation in areas like drug discovery and materials science.”
El Agente’s hierarchical agentic architecture enables intelligent and dynamic task distribution, automated error recovery, and continuous performance improvement based on past experience—all without requiring extensive user intervention. It also provides transparent action trace exports, supporting both reproducibility and human oversight. The system integrates a diverse suite of chemistry software and customized scripts, interfaces with high-performance computing job scheduling tools, and features a user-friendly interface with a chat window to interact with the agents and visualization tools.
“El Agente was created by a large multidisciplinary team of enthusiastic and talented researchers in record time,” says Varinia Bernales, director of research at the Matter Lab. “It showcases how collaboration allows innovation to unfold at a faster pace.”
Plans are also underway to integrate this agentic network into the development of self-driving labs (SDLs), an emerging technology at the centre of U of T’s Acceleration Consortium (AC), a global initiative accelerating scientific discovery. SDLs use artificial intelligence and automation to create new materials and molecules for a fraction of the usual time and cost. In this context, El Agente could be used for dataset generation, materials and process optimization, and safety, among other things.
By minimizing the technical barriers traditionally associated with computational chemistry, El Agente represents a step toward more inclusive, accessible, and scalable scientific research worldwide.
El Agente will soon be available to the chemistry and materials science community worldwide. To pre-register for the alpha and beta testing launch, visit elagente.ca and follow the Acceleration Consortium on Bluesky or LinkedIn.
