Designing an Agentic AI Assistant for Chemical Discovery

We employ a human-centered approach to design an agentic system that can address the unique challenges faced by chemists in the replacement of "forever chemicals". We conducted two formative studies with an aim to understand the current processes followed by chemists and the potential role of AI in augmenting and accelerating their scientific process. There’s a plethora of open source tools available for use that are specifically geared towards chemical and materials discovery and yet they pose a high barrier to entry for non-coding populations. Through the design of a system that not only allows access to such tools but also additional functionalities from generative AI to domain specific features, we hope to bridge the gap between subject matter experts and AI tools and functionalities that can help them. This work benefits chemists, scientists in domains like materials and drug discovery, toxicologists, regulators and other parties that form a multi-speciality team in tackling the global problem of forever chemicals.